Skip to main content

Full text of "Methods In Medical Research Volume 5"

See other formats


", -f'' 

I i \ ' 






t^gieK'r/itHiifijr.'Sfeif.feajgsf^StjC-'flaQagayyrgfc^Aw -- ■^' ''<£^s^jS3S3rta?8if=/M« 

Governing Bourd 

< 1 ?:, N. H. LONG 

EUGENE-A. ST?./».[? 







^ 'v 

V < », \ 

• * 







‘. V 









i V • i|. V, 

* ’-r Y 7^ ' . . V . n, m \ 

•;r.. - t* . « “■*£. ♦ • 'V', Jl K ^ 

■■5'A ■'* f'' - 'i.-'' '’A.'^ti * 

* it 

■. ^ 


' * 


* « 

r ■<^;. 




-.4, -f- 

.♦ »- 

»l ▼ 


Y'‘ ^ 

• >*. ^ ' 4 r • 

^ !W- k * 

' • *T 

» - « 

T ^ « < 

' >H 

iv >» 

^ I > 


H ■ • 



4 ,. * 1 ^, - 


*• -•# 

'' I '* 


' ^ ' V- ' 

. <1 ' 
r -» 7< 



AV * 
• • 

. « 

*% * 



IK* ^ 

’•ifc Xi 

• *»•■' - v '■ 

• ^ 41 * ^ 

« * 


>.-f' •■-' 

fcf : . * 

■ ■ * 1 

f%, ’* 

if • ‘ ^ 

^ ♦ Y 


%. • 

f; . - V*" 


^ ■* , .*- 


/f4*: V,* 

■- -w ••' ’•>. 

• » -V.'jt* 

•* • 

*▼ # 

f 1 ,' *' 


• , 

V t 

^ A 




ft' » « 

t i 

>• w* 


V x *i 

» • ' 

' '■■- ft 

' I?'** 

C v‘* , 

5 * #'. •* 

t* • 


> .,. 

* ■ V .. • .-. 

%*. • 

• '-,♦ t* 

« .V*\ » 


• 1 ' < 

* • • 


▼ ■ € 

• ' .* • 

V- “ 

' \ 

.. * 

/ “ 

* 4 




■• w 

- Ik • ^ . '^ ■' > ' >’ I S ' 

•V'^’ X'‘ ’ K> '■ 'i ^ '■*•' 

‘ ^ ' 1ft 'V 

k' 1 ^ K “ ' 

** 1 -.1 . T .-. , V* \*»r..,-.; ■• . 

t i: , 

•'V*' ( 

Volume 5 



VAN R. POTTER, Editor-in-Chief 

Assay of Antibiotics, Henry Welch, Editor 
Circulation—Blood Flow Measurement, Harold D. Green, Editor 
Selected Methods in Gastroenterologic Research, A. C. Ivy, Editvr 
Cellular Respiration, Van R. Potter, Editor 


JULIUS H. COMROE, Jr,, Editor-in-Chief 

Methods of Study of Bacterial Viruses, Mark H. Adams, Editor 
Pulmonary Function Tests, Julius H. Comroe, Jr., Editor 
Assay of Hormonal Secretions, Eleanor H. Venning, Editor 


RALPH W. GERARD, Editor-in-Chief 

Genetics of Micro-Organisms, S. E. Luria, Editor 
Assay of Neurohumors, J. H. Gaddum, Editor 
Selected Psychomotor Measurement Methods, Walter R. Miles, Editor 
Methods for Study of Peptide Structure, Choh Hao Li, Editor 


MAURICE B. VISSCHER, Editor-in-Chief 

llisTOCHEMiCAL STAINING METHODS, George Gomori, Editor 
Fluid and Electrolyte Distribution, Louis B. Flexner, Editor 
Studies on Gastrointestinal Pressures, Innervation and Secretions, 

J. P. Quigley, Editor 

Tissue Culture Methods, C. M. Pornerat, Editor 


Medical Research 


luviNE H. Page, Cfiairnian ; A, C. Ivy; Colin M. MacLeod; 
Carl F. Schmidt; Eugene A. Stead; David L. Thomson 

Volume 5 

A. C. CORCORAN, Edilor-in-Cldef 

WEIGHT SUBSTANCES, Lyman C. Craig, Editor 
METHODS OF RENAL STUDY, A. C. Corcoran, Editor 




^5 5 ? 







Methods in medic 

Printed in U.S.A. 


This series of volumes has now, we feel, become known and valued 
to the extent that it is no longer necessary either to defend its 
institution or to explain its purpose and plan. Dr. A. C. Cor¬ 
coran, who has been informally associated with the project from 
the beginning, now appears both as an Associate Editor and as 
Editor-in-Chief, and has furnished this volume with all the 
preface required. To him, to the other Associate Editors and the 
many contributors, we wish to express warmest thanks both on 
our own account and on behalf of the readers and users of the 
work. We are also indebted to readers of earlier volumes, and 
to other friends, for many helpful suggestions. 

It is with regret that the Governing Board accepts the resigna¬ 
tion of Dr. C. M. IMacLeod, whose other duties have pressed too 
heavily upon him. 

Irvine H. Page 
A. C. Ivy 
Colin M. MacLeod 
Carl F. Schmidt 
Eugene A. Stead 
David L. Thomson 

7 V 



■; i 


, ., fj 




'^. 4 tai 
!• '1 


The editorship of a volume of Methods carries a somewhat 
equivocal status. The only firmly established tradition is that 
the editor should furnish an introduction, even when it is his 
impression that it will not be read by most who use the book. 
The introduction has two major purposes: one, it should provide 
some background on the collation of the volume and, two, it 
should enable the Editor to justify his share of the task. 

At the outset, the Editor is guided in the selection of topics 
and the choice of Associate Editors by the Governing Board; 
the Associate Editors then order, arrange and edit their sections 
more or less autonomously. As a result, the Editor’s task and 
responsibilities are far from overwhelming. Dr. Cohn’s section 
on immunochemistry illustrates this principle of desuetude. For 
various reasons it was so delayed in transit that the Editor’s task 
was wholly undertaken by Dr. MacLeod, a member of the Govern¬ 
ing Board. To him our thanks and praise, as also the reader’s, 
for the material lies within his special competence and the out¬ 
come therefore more fortunate than the plan. 

Dr. Cohn’s section also demonstrates a major function of the 
Methods series. Those who seek acquaintance with immunologic 
methods will appreciate the scope and importance of its content; 
others, more familiar with the field, will recognize that this body 
of information could only otherwise be obtained by visits to the 
laboratories of the contributors or by tedious and often unsatisfy¬ 
ing searches of sources not readily available. 

Dr. Craig s section illustrates other aspects of modern scientific 
technology. The methods apply to a wide variety of molecular 
species and indissolubly embody both physical and chemical 
principles. It can be fairly presumed that the techniques de¬ 
scribed will be as useful in industrial as they are in medical and 
biologic research. Thus, it seems that both the disciplines and 
the orientations of science have begun to converge 

The section brought together by the Editor’s alter ego. Associate 
Mitor Corcoran, is certainly the most heterogeneous and least 
penetrating of the three. The procedures listed range from ultra- 
microscopic anatomy to routine clinical testing. This variety 
IS imposed by the topic, which is substantively the kidney as an 
orsan for study rather than as an isolatal)le aspect of fuimoral 




biology. Since the procedures of renal research are usually 
adapted or designed for purposes of a particular experiment, 
many of those described will serve more as exemplars than as 

Lastly, the Editor vishes to thank all who have contributed in 
any way and to join with Associate Editors and contributors in 
the hope that this volume—which is a tool, not a text—will find 
its place in laboratories rather than libraries. 

—A. C. Corcoran. 


BEYER, KARL H., M.D., Ph.D. 

Assistant Director of Research and Chief 
of Pharmacology Section, Research Divi¬ 
sion, Sharp and Dohme, Inc., West Point, 


Associate Professor of Medicine, College of 
Physicians and Surgeons, Columbia Uni¬ 
versity, New York. 


Institute for Medical Research, University 
of Louisville School of Medicine, Louis¬ 
ville, Ky. 


Fellow of the John Simon Guggenheim 
Memorial Foundation, and Research .4s- 
sociate, Department of Internal Medicine, 
University of Michigan Medical School, 
Ann Arbor. 

CLARK, J. K., M.D. 

Hospital of the University of Pennsylvania, 


National Research Council Merck Fellow, 
1949-5H; Institut Pasteur, Paris, France. 


Assistant Director, Research Division, and 
Staff Member, Cleveland Clinic Founda¬ 
tion, Cleveland. 


Member, The Rockefeller Institute for 
Medical Research, New York. 


Cytologist, National Cancer Institute, 
National Institutes of Health, Bethesda, 


Department of Physiology, New York 
University Bellevue Medical Center, New 


Associate Professor of Biochemistry, De¬ 
partment of Physiological Chemistry, 
University of IFtsconsin Medical School, 


Pharmacological Laboratory, University 
College, London, England. 


Director of Laboratories, St. Luke's Hos¬ 
pital, Cleveland. 

EARLE, D.VVID P., Jr., M.D. 

Associate Professor of Medicine, New 
York University College of Medicine, New 


Research Fellow in Surgery, Harvard 
Medical School, Boston. 


Professor of Zoology, and Chairman 
Science Division, Dartmouth College 
Hanover, N. H. 


Harold Brunn Institute for Cardiovascular 
Research, Mount Zion Hospital, San 


Director, Harold Brunn Institute for 
Cardiovascular Research, Mount Zion 
Hospital, San Francisco. 


Professor of Zoology, Syracuse University, 
Syracuse, N. I'.; present address, Ciba 
Pharmaceutical Products, Inc., Summit, 
N. J. 


Chief, Department of Cardiorespiratory 
Diseases, Army Medical Service Graduate 
School, .Army Medical Center, Washing¬ 
ton, D. C. 


Research .Associate, Boston General Hos¬ 
pital, Boston. 


Professor of Experimental Medicine, 
Southwestern Medical School of the 
University of Texas, Dallas. 


Assistant, The Rockefeller Institute for 
Medical Research, New Y'ork. 


Assistant Professor of Pediatrics, Western 
Reserve University School of Medicine, 

K.\B.\T, E. A., Ph.D. 

Associate Professor of Bacteriology, College 
of Physicians and Surgeons, Columbia 
University, New York. 


.Assistant Professor of Medicine, College of 
I hysxcians and Surgeons, Columbia Uni¬ 
versity, New York. 




Assistant Professor of Pathology, Western 
Reserve University School of Medicine, 


Staff Member, Research Division, Cleve¬ 
land Clinic Foundation, Cleveland. 


Research Associate, Institute for Medical 
Research, Cedars of Lebanon Hospital, Los 


Professor of Pharmacology and Associate 
Professor of Physiology, Bowman Gray 
School of Medicine, Walze Forest College, 
Winston-Salem, N. C. 


Member, The Rockefeller Institute for 
.Medical Research, New Y"ork. 


Member, The Rockefeller Institute for 
Medical Research, New York. 

MASSON, G. M. C., Ph.D. 

Staff Member, Research Division, Cleve¬ 
land Clinic Foundation, Cleveland. 


.-issistant Professor of Physiology, Uni¬ 
versity of Illinois College of Medicine, 


Assistant in the Department of Medicine, 
College of Physicians and Surgeons, 
Columbia University, New York. 


Assistant Staff Member, Research Divi¬ 
sion, Cleveland Clinic Foundation, Cleve¬ 


Professor of Chemistry, Harvard Medical 
School, Boston. 

OUDIN, JACQUES, Docteur en Mede- 
cine, Docteur es Sciences 
Chefde laboratoire, Institul Pasteur, Paris, 


Associate Professor of Microbiology, New 
York University College of Medicine, New 


Associate, The Rockefeller Institute for 
.Medical Research, New York. 


Specialized Instruments Company, Bel¬ 
mont, Calif. 


Professor of Physiology, Cornell University 
College of Medicine, New York. 


Assistant Director, Cardiovascular De¬ 
partment, Michael Reese Hospital, Chi¬ 


Department of Clinical Chemistry, Uni¬ 
versity of Edinburgh, Edinburgh, Scotland. 


Assista,nt Professor, Department of Internal 
Medicine, University of Michigan Medical 
School, Ann Arbor, 


Associate Professor of Bacteriology, College 
of Physicians and Surgeons, Columbia 
University, New York. 


Associate Professor of Physiology, Western 
Reserve University School of Medicine, 


Associate Member, The Rockefeller Institute 
for Medical Research, New York. 


Lilly Laboratory for Clinical Research, 
Indianapolis General Hospital, Indian¬ 
apolis, I nd. 


Research Assistant in Medicine, Mount 
Sinai Hospital, New York. 


Research Associate in Plant Physiology, 
University of Pennsylvania, Philadelphia. 


Welch Fellow in Internal Medicine, Na¬ 
tional Research Council; Assistant Pro¬ 
fessor of Medicine, College of Physicians 
and Surgeons, Columbia University, New 


Research Assistant in Medicine, Mount 
Sinai Hospital, New York; formerly 
Research Fellow in Experimental Medi¬ 
cine, Southwestern Medical School of the 
University of Texas, Dallas. 


Assistant Professor of Physiology, New 
York University College of Medicine, New 

WEST, CL.4RK D., M.D. 

Children's Hospital Research Foundation 


Director of Microbiological Chemistry 
Research Division, Sharp and Dohme, 
Inc., West Point, Pa. 


SECTION I. Methods for Separation of Complex 
Mixtures and Higher Molecular Weight Substances 



Countercurrent Distribution, by Lyman C. Craig. 

Paper Chromatography, by William Stepka (comment by 

Werner Hausmann). 

Electrophoresis, by Lewis G. Longsworth. 

Ultracentrifugation, by Edward G. Pickels. 






SECTION II. Methods of Renal Study 



Estimation of Renal Function from Plasma Disappearance 

1. Kinetics of Renal Excretion of Injected Substances, by 
Rafael Dominguez (comment by James S. Robson). 

11. Single Injection Technique in Evaluation of Renal Func¬ 
tion, by James S. Robson. 

Renal Blood Flow. 

I. Extraction and Clearance Method, by James F. Nickel 
and Stanley E. Bradley (comment by J. K. Clark) . 

II. Method for Direct Measurement, by Ewald E. Selkurt 

(comment by R. E. Shipley). 

Water and Electrolyte Metabolism. 

I. Infusion Technique for Measurement of Renal Function, 
by Norman Deane (comment by David P. Earle, 

II. Methods of Study of Body Water Compartments (Space 
Techniques), by Norman Deane (comment by David 
P. Earle, Jr., and Isidore S. Edelman). 

IIL Electrolyte Excretion Studies in the Dog, by Laurence 
G. Wesson, Jr. (comment by William A. Brodsky) 

IV. Determination of Water and Electrolyte Excretion dur¬ 
ing Osmotic Diuresis in Hydropenic Man, by William 














A. Brodsky (comment by Clark D. West and Laur¬ 
ence G; Wesson, Jr.). 

V. Bioassay of Antidiuretic and Diuretic Substances ■. . . 

Assay of Antidiuretic Substances in Urine, by J. Max¬ 
well Little. 

Concentration of Antidiuretic in Urine, by Louis B, 
Turner and Arthur Grollman. 

Assay of Diuretic Substances in Urine, by J. Maxwell 

Detection of Antidiuretic Substances in Blood, by 
Robert Gaunt (comment by J. Maxwell Little) 
Endogenous Creatinine Clearance, by Henry K. Schoch and 
Augusto A. Cdmara. 

Microbiologic Estimation of Arginine Tm in the Dog, by Karl 
H. Beyer and Lemuel D. Wright. 

In Vitro Methods for Study of Renal Tubular Excretion, by 
John V. Taggart and Roy P. Forster. 

Preparation of Tissue Section for Electron Microscopy, by 

A. J. Dalton (comment by Harrison Latta). 

Measurement of Renal Function in Rats. 

I. Measurement of Filtration Rate and TmpAH, by G. M. 
C. Masson and A. C. Corcoran (comment by Richard 

W. Lippman and S. E. Dicker). 

II. “Undisturbed” Method for Clearance Determination, 

by Richard W. Lippman. 

III. Determination of Inulin Clearance, by A. C. Corcoran . 

Experimental Renal Pathology. 

I. Measurement of Arterial Pressure in Rats. 

Microphonic Manometer for Indirect Determination of 
Systolic Blood Pressure, by Meyer Friedman 
and S. Charles Freed (comment by W. Glenn 


Recording of Indirect Blood Pressure in the Unanesthe¬ 
tized Rat, by Frederick Olmsted. 

II. Establishment of Hypertension in Rats. 

Hypertension Caused by Perinephritis, by G. M. C. 

Masson and A. C. Corcoran. 

Small Adjustable Renal Artery Clamp for Production of 
Chronic Systemic Hypertension, by Donald E. 


Hormonal Hypertension, by G. M. C. Masson and A. C. 


Saline Hypertension, by G. M. C. Masson and A. C. 
























III. Nephrotic Syndrome Induced by Injection of Anti- 
Kidney Serum, by Walter Heymann (comment by 
Abbie I. Knovvlton and Beatrice C. Seegal).... 264 

SECTION III. Immunochemical Methods for 
Determining Homogeneity of Proteins and Polysaccharides 



I. Production of Antibodies in Experimental Animals, by 

Melvin Cohn (comment by E. A. Kabat) .... 271 

II. Separation of Antibody-Active Proteins from Various 
Animal Sera by Ethanol Fractionation Techniques, 

by Harold F. Deutsch.284 

III. Techniques and Analysis of the Quantitative Precipitin 


A. Reaction in Liquid Media, by Melvin Cohn (comment 

by E. A. Kabat).301 

B. Specific Precipitation in Gels and Its Application to 

Immunochemical Analysis, by J. Oudin.335 



Owing to the length and special production problems of this volume, 
royalties are being applied to the costs in order to maintain the sell¬ 
ing price at the lowest possible level. Royalties for all other volumes 
of the series are distributed among the Editors and Associate Editors. 


Methods for Separation . 
of Complex Mixtures and 
Higher Molecular Weight 



IMedical research has always profited from advances in the 
understanding of the complex physicochemical systems found in 
living matter. Among the techniques needed, not the least impor¬ 
tant are those which permit the separation and recovery of crucial 
substances from the biochemical matrix, occurring as they often 
do in relatively small amounts. 

A large number of special methods are available for the deter¬ 
mination or isolation of specific substances, but among the general 
methods of organic chemistry, such as distillation or crystalliza¬ 
tion, many are not suitable, either because of the unstable nature 
of the biochemically interesting compounds or because of the 
initial complexity of the mixtures. 

Extraction with solvents has always been one of the most useful 
general procedures. As countercurrent distribution, it has now 
been advanced to the position of a highly powerful and specific 
preparative and analytical approach which can be adapted to a 

wide variety of substances, especially those requiring mild condi¬ 

Similarly, a/lsorptlon techniques have long been among the most 
useful tools for separating substances of biochemical interest. 



Paper chromatography is a new development in this field which 
has already firmly established itself as one of the outstanding ad¬ 
vances in the biochemical methods of recent times. 

When larger molecular species are involved electrophoresis and 
the ultracentrifuge have been accepted without question for more 
than a decade as necessary tools. Nonetheless, recent years have 
brought forth improvements in technique for both methods. 

— J. Dklafield Gregory and Lyman C. Craig. 


LYMAN C. CRAIG, The Rockefeller Institute for Medical Research 

Nearly everyone working in tlie biological sciences as well as 
in chemistry is familiar with simple extraction as a tool generally 
applied in the first preliminary investigations of a new problem. 
It has been used this way for the separation of substances with 
widely differing properties since the beginning of chemistry. Only 
in recent years have its possibilities as well for the final character¬ 
ization of a wide variety of closely related substances been really 
appreciated. Naturally realization of the fullest potentialities can 
be accomplished only through its integration with the many useful 
and specialized analytical techniques now available, and by rather 
extensive equipment. However, even with the techniques and 
equipment of most laboratories, far more can be accomplished than 
has been done in the past. 

In dealing with an unknown preparation one of the first con¬ 
siderations should be that of fractionation. A mixture requires 
fractionation for isolation of the component of possible interest. 
A supposedly pure substance also requires a fractionation attempt 
in order to prove that it is not a mixture. 

All methods of fractionation, in the broadest sense, have a 
common theoretical basis in that they deal with the manipulation 
of fractional parts. To be sure, in certain cases, the fraction re¬ 
moved or transferred is so large that for practical purposes separa¬ 
tion is complete. The majority of separations, however, are not 
accomplished so easily and losses occur at each step, losses in 
labor as well as in material. Proper systematization minimizes such 
losses. This can be understood perhaps better by stepwise frac¬ 
tional extraction than by any other approach. 

* us take the case of a single solute, A, which can be dis¬ 
tributed between 2 immiscible solvents. If at equilibrium with 
equal volumes of the 2 phases, half of the solute is found in each 
phase then the partition ratio of the solute is 1. If the upper phase is 
transferred to a new equilibration tube (separatory funnel) con¬ 
fining an equal volume of fresh lower phase, and an equal volume 

fresh uiyer phase is added to the first tube, the state of affairs 

total solute will be m each of the 2 units. Now if both are equili- 

‘“be 2, that from tube 

u transferred to tube 1, fresh upper phase added to 0 and fresh lo«v 





phase added to tube 2, the state of affairs shown by line 2 will be 
reached: 0.25 of the original will be in tube 2, 0.5 will be in tube 1 
and 0.25 will be in tube 0. If this process of alternate equilibration 
and transfer is continued until a total of 9 units is in the series, then 
the distribution of solute in each unit will be shown by the bottom 
line of the table. The number of stages applied is given by the 










































































Let X = Fraction in 
upper layer 

Let y = Fraction in 
lower layer 

Fig. 1.—Distribution series; 8 stages. 

vertical column on the left while the number of the unit is given by 
the top horizontal column. 

If the fraction of the original in a unit is plotted against the con- 
.secutive number of the unit as given in the table, a distiibution 
curve can be obtained with a maximum at tul>e 4. Had the parti¬ 
tion ratio been 3, then the distribution would have been the cuive 
on the right for 8 stages, while for a ratio of 0.333, it would have 
been that on the left. If 2 solutes had been present in equal amount 


iuul tlie ICs had been 3 and 0.333, respectively, then the separation 
in each tube would be plainly evident from the 2 curves. 

xMathematically, the Table of Figure 1 is represented by the 
binomial expansion, (X + F)” = 1, where n is the number of stages 
or transfers, X is the fraction in the upper phase and Y is the trac¬ 
tion in the lower phase. If X is the fraction in the upper phase of a 
single tube, then 1 - is the fraction in the lower phase. By defi¬ 
nition, the partition ratio K is X /(1 — X). Thus X is equal to K/- 
{K -h 1) and the fraction in the lower phase is 1 — [K/{K H- 1)] 
or 1/(1 K). The binomial expansion in terms of the partition 

ratio for equal volumes becomes 

\K + 1 



1+ A' 

( 1 ) 

Obviously it soon becomes laborious to perform such multiple 
extractions individually in separatory funnels as represented. 
Nonetheless many laboratories have performed separations in¬ 
volving up to 20 or more separatory funnels in the series. Fortu¬ 
nately, mechanical equipment has been developed (10, 12, 20, 
30, 16, 29) for the purpose and the number of units in the series 
can now be expanded almost indefinitely if the particular fractiona¬ 
tion problem requires it. One extractor in use at the Rockefeller 
Institute (10) contains 220 equilibration units. It is fully automatic 
and is equipped with electric motors, time clock, etc. From 100,000 
to 200,000 individual extractions can be performed in a 24 hr 
period. This equipment is commercially available.* 

Such mechanical equipment, however, is valueless unless suitable 
systems can be found. The term system as used here means any 
mixture of liquids or solids which will yield 2 clear liquid phases 
when they are brought together and thoroughly mixed. The 2 
phases are always equilibrated at the temperature of the experi¬ 
ment just prior to their use. To be suitable, a system must meet the 
following main requirements. 

1. It must be capable of dis.solving the solute of interest and 
should furnish a partition ratio within a practical range. If ecjual 
V ohimes of each phase are to be used in the mechanically operated 
e(iuipment, the range would be from 0.01 to 100. With hand- 
operated equipment the range should be much narrower, from 0.2 
to 5. In the latter case, however, compensation can be made easily 
tor a A too high or too low by using a larger or .smaller volume 
m ei her the upper or lower phase, dims where /•„ is the i-atio of the 
\ olumes, the product Kr„ should be within the limits of 0.2-5. 

1 o reach the best separation possible with the least number of 

* Obtainable from 0. Post. G822-00th Road, Maspeth. N. Y. 



transfers, with a mixture of 2 components, the geometric mean 
of the 2 products, Kn and K'vy should be 1 (6). Here K and K' 
are the respective partition ratios of the 2 components. 

2. The partition ratio must be reasonably independent of the 
concentration. All solutes and systems show more or less deviation 
of K with concentration, particularly at higher concentrations. 
The adherence to ideality required will vary with the number of 
transfers applied. Thus a variation of 10 per cent over a 100-fold 
concentration range is not serious if no more than 20-30 transfers 

Fig. 2.—Test tube rack for preliminary countercurrent distribution. 

are to be applied. But, when hundreds of transfers are involved, 
closer adherence is desirable. 

In dealing with a new solute it is always worth taking the time, 
before launching forth on a countercurrent distribution run, to 
determine the K at the maximum concentration initially to be 
used in the run and at Vio this concentration. 

3. The system must permit equilibrium to be reached rapidly. 
This point is easy to test experimentally (3) when the partition 
ratio is determined. In general, partially miscible solvents cause 

little trouble in this respect. • . u 

4. The phases must separate clearly within a few minutes alter 

equilibration and must not form stable emulsions, . 

tend to form, lowering the pH probably will help. Addition of a 


trace of a salt often is helpful. Increasing the volume of the organic 
phase often promotes separation. On close examination it may be 
found that a small amount of a solid film is separating at the 
interface. Removal of this will promote separation. 

5. The solute must not be unstable in the system. 

6. The system must lend itself readily to quantitative solute 
analysis, preferably by direct weight determination and to con¬ 
venient recovery of the solute after the distribution is finished. 

7. The system should not be expensive, extremely toxic or 
corrosive to the equipment. 

Other less important points could also be mentioned. A system 
deficient in 1 requisite may be so favorable in the others that it is 
worth using in spite of disadvantages. 


When a suitable system has been chosen a preliminary run can 
be made. It may be made in the distribution apparatus or pref¬ 
erably in a row of glass-stoppered test tubes supported on a rack 
(12) as shown in Figure 2. The apparatus of Figure 2 can be 
assembled from standard laboratory equipment. 

The tubes are held by spring clips attached by a screw to a short 
piece of Flexaframe rod. The stainless steel Gee clips, available from 
scientific supply houses, may be conveniently adapted. The rods 
attached to the clips are in turn attached to a longer rod which 
extends thiough 2 bearings on each end of the rod. The bearings 
aie simply Flexaframe clamps in which the screw remains loose. 
The bearings are supported by 2 stands. A crank is attached to 1 
end of the rod for tumbling the tubes. 

In this design each tube receives its equal portion of stationary 
phase initially. The sample to be distributed is placed in tube 6. 

portion of the other phase, previously equilibrated 
with the hrst, is then introduced and equilibrium is reached by 
inverting the tubes with the crank. Twenty-five inversions are 
usually sufficient (.3). The glass stopper will stick sufficiently not to 
fall out, if It IS given a slight turn when it is placed tightly in posi¬ 
tion. Small interchangeable stoppers are therefore more reliable 
than larger ones. 

fpAnT H® 'T''-’' fphase to be moved is trans- 
feired to the adjoining tube, 1, by a small siphon (Fig, 2, inset) pref- 

htig”l®ert7 thf rubber collar, 

hot; tfcZriT PassefZough 

collai and has a hole at the bend. The operator’s finger 



is placed over this hole, so that instant release of the vacuum or 
pressure is possible. 

Following the first transfer, fresh mobile phase is added to 0 
and an equilibration made. The mobile phase of 1 then goes to 2, 
that of 0 goes to 1, and fresh phase goes to 0. This procedure is con¬ 
tinued until all the tubes of the series contain both phases; i.e., 
the fundamental operation is complete. A series of 12 tubes can be 
filled in this manner in approximately 1 hr. 

Analysis of the solute content/ml in both phases in each tube is 
then made and the result is plotted as in Figure 1 except that 
weight or some analytical unit proportional to weight/ml is used 

3—Thirty tube hand-operated train. 

as the ordinate. Two curves will be obtained, 1 for the upper 
phases and 1 for the lower. From the data the partition ratio in 
each tube can be calculated directly. With a single solute or mix¬ 
ture with quite similar partition ratios the determined K m each 
tube throughout the series will be constant and the upper and lower 
curves will differ only in height rather than in position. 

The pattern may reveal that a mixture has been sharply sepa- 
rated into 2 or 3 separate bands. More likely, however, a mixture 
will be revealed by a progressive increase in K from the lowest to 
the highest tube. Here a more extensive run with a machine is 
indicated. If so, the contents of each tube can be transferred to the 
corresponding tube of the machine in order to continue without 

making a fresh start. 


The type of machine at present employed in the author’s labora¬ 
tory is actually a glass train (Fig. 3). The shorter trains with 25- 
100 units arc hand-operated. A mechanical robot has been con- 

Fig. 4.—.Automatic 220 tube apparatus. 

k's a zzu Uit)e 

structed for operating the longer trains. Figure 4 show. 

tram equipped with sueli a robot. 

■; Tht “"fof the individual tube or cell is shown in Fiaii re 

32lm i::" ThTsluhrat'f - “PP-imS; 

The decantation tube c is W cm Iona 11^^™ ’’ll"'" 

wall of the decantation compartmentd aFaTig^StlStlrds 



nearly to the opposite wall. A shallow enlargement is blown in d 
opposite the opening of c. Like the cell, d is made from 15 mm tub¬ 
ing. Its opening at e is either sealed to the next equilibration cell of 
the series or is joined to it by a flat ground-glass stopper. 

Access to each equilibration cell is permitted through h, which is 
a flat ground-glass stopper designed to close the tube /. The 
opening tube / is 4 cm long on the present apparatus. However, 
experience would now seem to suggest that a length of 8 cm would 
be preferable. The stopper h is held in position by a simple spring 
clamp D. Any device for opening and closing/must be simple and 
extremely rapid. Otherwise with 220 tubes in the series excessive 
time is required for the opening or closing operation. 

The 2 phases in the equilibration cell are brought to equilibrium 
by rocking from position A to B and back again. Each position is 
about 40° from the horizontal. After a predetermined number of 
strokes, 5-50, depending on the rate at which equilibrium is estab¬ 
lished (3), the phases are permitted to separate at position B. 
Upon tilting further to position C, the upper phase decants through 
c to d. The length of the lower part of the cell is adjusted at the time 
of construction so that with a 10 ml lower phase, the meniscus 
separating the 2 phases only reaches point a. When the cell is tilted 
back to position A, the contents of d flow through e into the next 

adjoining cell of the series. 

The cells are supported on two 4 X 4 cm duraluminum bars, 
each 180 cm long (Fig. 4). Two rows of cells are thus provided. 
The upper row is as close to the lower row as the height of the 
decantation tube will permit, but the former is placed slightly to the 
rear so that the stoppers closing the lower tubes are readily accessi- 

ble. , 

The cells are held on the metal bar by means of a thin metal 

strap of stainless steel which passes over the glass tube and presses 
it against the duraluminum bar by a small bolt. The bolt passes 
through the bar to a nut on the under side. Notches have been cut 
in the bar so spaced as to receive each cell and hold it in position. 

In the apparatus currently in use all the cells are sealed togethei 
through each overflow tube, e, except every lOth cell. At this 
interval they are joined by flat glass joints. The cells can therefore 
be handled and removed from the bar in banks of 10. Less oppoi- 
tunity for leakage is thus presented and it is necessary to use only 
2 metal straps for the support of the 10 tubes to the duraluminum 
bar On more recent commercial models each cell carries a joint 
and can easily be replaced in case of breakage. For a more complete 

description of the apparatus see reference (10). ^ qmvrpnt 

In studies with the distribution machine thus far, o diflerent 


methods of operation have proved most useful: (1) fundamental, 
(2) single withdrawal, (3) alternate withdrawal, (4) completion of 
squares, and (5) recycling. 


This procedure must always be employed even when 1 or more 
. of the other procedures is to be used later. The last 4 procedures are 
thus supplementary and permit a higher order of separation to be 
achieved without increasing the number of tubes to be manipulated 
in the series. 

In the fujidamental procedure, somewhat in excess of the amount 
of lower phase required for the entire train is inserted into the 
apparatus. The excess solution flows along as the run develops 
and finally emerges at the last tube. 

After tipping the train to the decantation position a number of 
times, the solute to be distributed is inserted into the first tube with 
the upper phase (or into a small number of the first tubes). Several 
portions of upper phase are then placed in the tubes in front of the 
solute. These condition the lower phases before the solute reaches 

A transfer is then accomplished and a fresh portion of upper 
phase is inserted at the 0 tube. The filling is made automatically if 
a filling device (10) is provided. The process is repeated until all 
the tubes contain their complement of upper and lower phase. 
Analysis is then made and a distribution pattern is plotted. This 
completes the fundamental procedure. 

In case the pattern shows insufficient separation, an intelligent 
choice of one of the supplementary procedures can be made. Here 
the use of theoretical curves for comparison to those of the experi¬ 
mental will be helpful. The calculation of such curves will be taken 
up later. 


If a spectrum of partially separated bands is revealed by the 

undamental pattern, the run can be continued by permitting the 

ast upper phase to flow from the series at each transfer Here 

tbe outgoing phase is assigned the transfer number which caused 

It to emerge from the series. It is set aside to be analyzed The 

fraction collector of the automatic apparatus (10) takes care of the 

tn automatically. The series emerging is analogous 

to that of effluent chromatography. ^ 

After an arbitrary time tlie run is interrupted for analysis again 
A sufficient number of the tubes of the train and of the eE 
Mines are analyzed so that distribution patterns can be plotted 



The pattern for the effluent series is placed to the right of the 
fundamental series with the highest transfer number adjacent to 
the highest tube number of the fundamental series. The abscissa 
for the effluent series is in terms of the transfer numbers which 
have been assigned to the effluent fractions. Again the pattern 
obtained is interpreted by the use of theoretical curves. 


In case the fundamental pattern shows that material in 1 or 
more of the earlier tubes of the series has been adequately frac¬ 
tionated, the contents of these tubes can be withdrawn and re¬ 
placed by fresh solvent. Withdrawal can be made by removing a 
whole bank of tubes at 1 time or by removing a tube on every 2d, 
3d, 4th, etc., transfer. Selected tubes are likewise removed from the 
other end of the series which contain the most rapidly moving sol¬ 
ute. Two effluent series and a fundamental series thus can be 
obtained. One effluent series contains solute with lower K while 
the other contains solute with the highest K. The fundamental 
pattern contains solute with intermediate partition ratios. In 
plotting the patterns, the result for the material of lowest K is 
placed on the left of the fundamental pattern; that for the higher 
K on the right. Alternate withdrawal is most conveniently carried 
out with a circular apparatus (described in reference (12)) or in the 
glass train (described in reference (10)). A more complete de¬ 
scription of the procedure is given in reference (12). 


In this procedure, which is a systematization devised by Bush 
and Densen (6), the fundamental series is completed and the dis¬ 
tribution is then continued without addition of upper phase. The 
effluent upper phases are collected as in single withdrawal. When 
the last upper phase has left the train 2 distribution patterns are 
constructed from the analytical data obtained from the effluent 
and the lower phases remaining in the train. The data are plotted 
in the same manner as in single withdrawal except that the pattern 
of the fundamental series is replaced by that of the lower phases. 
The tube numbers which were the abscissa of the fundamental 
pattern are therefore replaced by transfer numbers, with assign¬ 
ment to each tube of the transfer number which caused the upper 
phase to migrate from it. 


At the point of completion of the fundamental pattern, 2 or 3 
overlapping bands may be revealed with no solute in either or both 


extremities of the pattern. More transfers may be applied by re¬ 
introducing the effluent upper phase from the highest tube as the 
upper phase required at tube 0. This can be continued until the 
front of file advancing solute band almost catches up with the 
trailing edge. In this way many rhore transfers may be applied and 
a much better separation achieved without greatly increasing the 
total volume of solvent. 

Such a procedure becomes most useful when a considerable 
number of units or tubes is in the series. For example, with a single 
pure solute being distributed in a train of 25 tubes, the advancing 
solute front would already be overtaking the trailing rear at 60 
transfers. On the other hand, if 220 tubes are in the series the 
advancing solute front would not be overtaking the trailing edge 
until more than 5,000 transfers had been applied. 

To take full advantage of this fortunate circumstance, the auto¬ 
matic 220 tube apparatus (Fig. 4) is arranged so that it really is a 
double train containing 110 cells in each series. In the lower series 
the upper phases flow in the opposite direction to those in the 
upper series. Thus tube 219 is directly under tube 0 and the upper 
effluent phase can be diverted automatically into 0 again if this is 

In this procedure the results are plotted as in the fundamental 
case provided solute has not migrated beyond tube 219. Where it 
has re-entered the series for a second circuit, tube 0 becomes 219 
+ 1 or tube 220. Tube 1 becomes 221, etc. 

Calculation of theoretical curves .—A single pure solute behaving 
ideally in a given system will give a characteristic and reproducible 
distribution of mass throughout the series of cells. Since this dis- 
tiibution will depend only on the number of transfers applied and 
the partition ratio (or Kr, where unequal volumes of the phases are 
used), the di^ribution to be expected can be calculated directly 
at any time. Comparison of the calculated curve with the experi¬ 
mental one then permits a much more critical evaluation of the 
question of purity or, if the pattern indicates overlapping bands 
theoretical curves permit the most intelligent choice of one of the 


viously all of these cannot be treated hl beca^^ ;! the iL^’ 


^ From the binomial of equation (1) the fmctinn , solute present in any one hlbt’ (in np;:":ncn,:l;:r”;S:s 



combined) can be calculated. A general formula for this purpose 
(31) is found in equation (2) where 

= (2) 

is the fraction in the rth tube at n transfers. Rather than calculate 
each value independently by equation (2), the desired curve can 
be reached more rapidly by calculating a single term first by equa¬ 
tion (2) and then by calculating the remainder from this term 
through the use of simple factors as given below. 

In general the rth term is related to the (r l)th term by equa¬ 
tion (3), where F' = {r l)/(n — r). The F' factors are thus 1/8, 
2/7, 3/6, ....8/1 

Tr+i = TrK/F' (3) 

for calculation of the 1, 2, 3, .... 8 tubes from tube 0 in an 8 stage 
distribution. The fraction in tube 0 is {\/K + 1)). Similarly, the 
factors for a 10 stage distribution would be 1/10, 2/9, 3/8, .... 
10 / 1 . 

When higher numbers of transfers are concerned and where the 
band has cleared both the 0 tube and the highest tube, a different 
method of calculation is preferred. The approximation in equation 


y = _ ^ - e-xy[2nK/{K + 1)»] (4) 

\^2nirK/{K -f- 1)* 

which can be derived from the equation for the normal curve of 
error, permits a single curve for any number of transfers n to be 
calculated with the aid of a slide rule in 10 min. 

In matching a theoretical curve to the experimental, the yo 
term of the distribution usually is assumed to be the maximum of 
the experimental curve. Since here x is 0, equation (4) simplifies to 
equation (5). 

__ 1 _ ( 5 ) 

“ V2n^K/{K + IP 

I^ollowing the assumption that yo is tlie experimental maximum, 
the next point of interest becomes the degree to which the othei 
calculated points of the theoretical curve will match with the 
experimental. Tlie calculated y values each are related to yo by the 


g-xVl2«K/(K + 1)*1 

Therefore 2nK/{K -f 1)^ can be considered as a constant appearing 
in the calculation of each Y value. The constant e also appears in 



each calculation. In calculating a ij value x tubes removed from yo 


2nK/{K + 1 ) 

is the logarithm to be used as a factor in each calculation. Let this 
logarithm be L. The fraction of the original solute present in any 
tube X tubes removed from yo is therefore related to t/o by the 
factor 1/antilog of x^L. Calculation of approximately 5 of these 
factors is sufficient for a theoretical curve to be matched. 

In the case of a single ivithdrawal series the mathematics are 
even less complicated than in the case of the fundamental series. 
Here the partition ratio can be calculated from equation (6) 

K = u/(n — u) 

( 6 ) 

where u is the number of tubes in the train and n is the transfer 
number of the maximum. The theoretical curve can be calcu¬ 
lated as given above for the fundamental series except that equa¬ 
tion (7) 



must be used instead of equation (4). 

Calculations for the alternate withdrawal series are much more 
complicated and will not be treated here. However, there is little 
difficulty in calculating curves for the completion of squares pro¬ 
cedure or for the recycling procedure. The latter is identical with 
the calculation of the fundamental curve. Bush and Densen (6) 
have given excellent mathematics for calculating the former. 
Where many tubes have been used in the series, equation (7) 
cmi be employed for calculating the theoretical curve for the 
effluent upper phase. The lower phase pattern can be considered 
^ the inverse of this. Therefore l/K should be used in equation 
(7) for calculating the theoretical for the effluent lower phases K 
in this case is obtained from equation (8) 

K = s/{n — s) 

( 8 ) 

where s is the number of lower phases left without 

determinat'ion of weight patterns 

—--- OiliUl ^ . 


a) Apparatus. A series of thin glass shells serve both as 
evaporation and weighing vessels. They are made by blowing a 
round bulb, approximately 3.3 cm in diameter, from soft glass 
ampules. The lower half of the bulb is detached by scratching a line 
on the fragile wall with a diamond point and then cracking off the 
hemisphere with the hot wire of a glass cutter. The cracked edges 
are strengthened by fire polishing. A sufficient number are made so 

that shells weighing within ±50 mg of each 
other can be selected. The weight of a finished 
shell should approximate 0.5 g. 

The shells are arranged in order of increas¬ 
ing weight on a wire frame (Fig. 6) which can 
be made by twisting a series of loops from a 
copper wire. Loops of the proper size result 
from molding the wire around a glass tube of 3 
cm OD. The wure frame is hung on a small tri¬ 
pod which also supports a pair of small forceps 
for manipulating the shells. The jaws of the 
forceps can be covered with small pieces of rub¬ 
ber tubing to avoid damaging the fragile shells. 

The evaporator is a small steam bath placed 
on an electric hot plate. The cover of the steam 
bath is fashioned from a flat sheet of stainless 
steel with 5 holes drilled in it. One of the holes 
permits a reflux condenser to be attached, and 
the other 4 are of such size, 3.0 cm, that the 
glass shells will not fall through, yet large 
enough so that most of the under surface of the 
shells will be exposed to the hot vapors when the 
shells rest in the holes. A glass tube, 10 mm ID, 
with its opening approximately 2 cm above 
each shell serves as an air jet. 

The top shell of the series is a control tare. If 
the solvent system itself contains a measurable 
residue, it may be advisable to evaporate an ap¬ 
propriate aliquot of the solvent as a blank each 
time a series of determinations is made. In this 
manner the weight of each fraction subsequently determined is 
automatically corrected. Otherwise, the tare only receives the same 
final drying treatment as the other members of the series. 

b) 'Minique .—An aliiiuot of the solution to be analyzed, 0.1- 
3.0 ml, is added to the second shell after it has been placed on the 
steam bath. A hypodermic syringe is ideal for this purpose, and can 
be fitted with a fine glass tip drawn out from an adapter, t A current 



of air to the jet is filtered free of dust particles through a cotton 
plug and is blown at the solution with no more than sufficient 
speed to be barely noticeable on the surface of the solution. 

If an aqueous solution is to be analyzed, it will require 2-3 
min for a 1 ml aliquot to be brought to dryness at 100°. Higher 
or lower temperatures can be reached as desired by placing liquids 
of suitable boiling temperature in the bath. During the time re¬ 
quired for the evaporation of the first aliquot, others can be started 
on the remaining 3 holes of the bath. 

As soon as the solution comes to dryness, the liquid clinging 
to the bottom of the shell is touched off with cotton gauze, and the 
shell is placed in its proper position on the wire frame. When all of 
the shells have been so treated, the wire frame with its shells is 
placed in a large glass test tube fitted with a rubber collar near its 
upper end. The tube and its contents are supported in a steam bath 
by means of this collar, and the open end is closed by a I-hole 
rubber stopper with a glass tube and rubber hose eonnection to a 
high vacuum oil pump. All the samples and the tare are dried to¬ 
gether at I00°/0.2 mm for the desired time, usually about 5 min. 
The shells are then weighed on a semimierobalance to ±0.01 mg. 

In weighing, the tare always counterbalances the shell containing 
the residue. 

^^^d^^'^r-Kohlbusch balance used in this laboratory utilizes 
riders for weights up to 100 mg, and hundredths of a milligram are 
obtained by measurement of the degree of defleetion. Since the 
shells do not differ from the tare by more than 100 mg. the entire 
series of shells can be weighed by use of the riders alone. The 
balance IS magnetically damped, and each weighing requires no 
longer than IV 2 mm. 

Systems for separating various types 0 / compounds.—The decision 
as to whether a particular system is the one most satisfactory for a 
given type of substance can be very time-consuming if the separa¬ 
tion IS likely to be difficult and if many different combinations of 
so vents are tested. Much time and efihrt have ber^ent along 

more ti’mJ I" laboratory, perhaps coMiderably 

moie time than has been devoted to the theoretieal, mechanic^ 

tributfon^irtherer*' development of countercurrent dis^ 
bution. It theiefore appears highly advisable to mention some 

~ -ed f:~ 

dic”a:w:Tsvl‘.lr,'°'';" --ds and 


t Available from Bect„„,Dicki„aoi..„d Co., Ruthertord.N.J. 



solvents for the nonpolar phase has been most useful. The organic 
phase can be isopropyl ether, chloroform, benzene, heptane or an 
alcohol such as butanol. Isopropyl ether redistilled and preserved 
over iron wire to retard peroxide formation has been one of the 
most useful solvents for the organic phase. The desired value of 
partition ratio can be obtained by varying the pH and concentra¬ 
tion of the butfer or by choosing an organic solvent with more 
hydrophilic properties where the K favors the aqueous phase. 

The reverse is followed where the solute favors the organic phase. 
Mixtures of solvents can also be useful for reaching a satisfactory 

If an example, Figure 7 sliows the separation of the C., 

Cfi C 7 and Cg acids in 52 transfers. For the higher fatty acids hm 
ever, a buffered system did not prove suitable because of ^en 
poor hydrophilic properties and tendencies to give f " ’ 

Here a system made by equilibrating equal Pf*';* 
glacial acetic acid and formamide appeared satisfactory, g 


shows the separation obtained in 400 transfers with a mixture of 
300 mg each of C 12 , Ch, Cie and Cis acids (I). 

ORGANIC BASES. —The type of system found most useful for bases 
is very similar to that most selective for the organic acids. In fact, 
buffered systems (9, 11, 14, 28) gave relatively much greater selec¬ 
tivity in separating basic mixtures. The method is ideally suited 
for the separation of complex mixtures of alkaloids (13). In the 
majority of cases the buffered systems are so selective that dis¬ 
tributions involving high numbers of transfers are not at all 

AMPHOTERIC SUBSTANCES. —Ill this class of substaiice much work 
remains to be done. Buffers appear to contribute little to selec- 

Fig. 8.— Separation of C,,, Cu, Ci, and Cis acids. 

tivity. Incorporation of highly charged solutes such as salts into the 
system appears to give somewhat greater selectivity, but here the 
problem 0 ehmmat.on of the salt in order to permii weight mmh 
ys.s partm ly offsets the advantage. Ammonium acetate does not 
ave a higl, saltmg out (or salting in) effect but can be elimTnated 
volatility. It is therefore a valuable adjunct If the 



TABLE 1.—Partition 

Ornithine • HCl 















Arginine • HCl 














. 15 







Ratios of Amino Acids 

Aspartic acid 






Glutamic acid 



a-Aminobutyric acid 


















Phenylah’ ine 



Tryptopl n 



hydrochloric acid. Most of these values have been substantiated 
in actual runs (10). 

Where bands overlap in this acid system they may be separated 
in a neutral n-butanol/water system or in an alkaline system. A 
system made by equilibrating a solution of 2 parts sec-butanol and 
1 part propanol with an aqueous solution containing 10 per cent 





c, 10 



6 0.0 









1 « 






r - 



Tyrocidine A Hydpolysat< 
*—* Weight -lower 
• “ -upper 

*—^ Proline by colour 
0 —-o Theoretical 






80 160 Z 

Tube No. Tpanster No. 

9 —Sepuration of amino acids from tyrocidine A. 

ammonium acetate and 3 per cent ammonia permitted the separa¬ 
tion of phenylalanine, leucine, tyrosine, valine prohne and a 
band containing aspartic acid, glutamic acid and ornithine. 1 he 
pattern shown in Figure 9 was obtained on a hydrolysate ot tyio- 

’’llv thiseparation of the higher polypeptides a variety of systems 



have been developed. Table 2 gives a few of the systems most useful 
for the various classes of naturally occurring peptides thus far 
studied in our laboratory. Although insulin properly belongs to 
the protein class, it too has been brought under study and success¬ 
fully distributed. 

Substances which are so strongly hydrophilic that they are 
scarcely soluble in any organic phase can be complexed with a 
suitable comple.xing agent such as a detergent. These agents have 
l)een called carriers by O’Keefe et al. (23). This approach has been 

T.\BIiE 2.—Systems for Peptides 



Gramicidin-S peptides 
Polymyxin type 

Ethyl ether/phosphate buffers 
System made from water, methanol, chloroform 
and benzene in 7, 28, 15 and 15 vol proportions, 

Methanol, chloroform and O.IN hydrochloric acid 
in 2, 2, 1 vol projiortions, respectively 
Methanol, chloroform and 0.01 N hydrochloric 
acid in 2, 2, I vol proportions, respectivelv 
Systems made with 2-butanol and 0.01-0.5N hy¬ 
drochloric acid 

2-l)utanol/3% aqueous acetic acid or n-butanol/ 

2-butanol/l% aqueous dichloracetic acid 

useful in developing systems for streptomycin (27), neomycin 
(26), heparin (22), nucleotides and phosphorylated sugars (24). 
I he distribution of insulin (17) using di- and trichloracetic acid 
in the system probably also comes in this category. 

A number of interesting systems for separating sterols have been 
studied by Cornish et al. (7). Heptane and acetonitrile form a par¬ 
ticularly good system. The chlorophyls can be separated readily i 
the systern hexane/90 per cent methanol. Golumbic and' 
workers (15) have given svstems for 

separating phenolic 


dif hydrochloride. Since gramici- 

din-S was known to contain only basic groups it was thought bLf to r 

ribute it m a system containing hydrochloric acid. This would maintain it 

f s'- ‘-in (12) oontain- 
perature of the room sufficient of the he" ^nfl eqml'brating at the tem- 
lower phases in eaeh/,t™fs::tte rthetfl 
Hie apparatus was then tipped to the def-nUin f- ^ ^ 

thus clearing the fi.t lo tubes of exces., lower moving tt*^: 



empty tubes further along the train. Volumes of the lower phases are auto¬ 
matically adjusted in this way as the distribution proceeds. 

Ten upper phases were then placed in tubes 3-13. These phases serve as a 
forerun to insure proper equilibration of the system. The sample 1.0 g, was 
placed in 30 ml of each phase and these solutions divided among tubes 0,1 
and 2. 

Equilibration was accomplished by rocking back and forth 10 times. 
The phases w'ere permitted to separate with the apparatus tipped halfway 

to the decantation position, then tipped farther so that all the uppei 
phases decanted and then tipped back again to the transfer position. A 
fresh upper phase was added to the 0 tube, etc. After 106 transfers the ef¬ 
fluent was collected in test tubes marked with the transfer number on 

which each fraction emerged from the train. 

At 340 transfers the distribution was evaluated by weight analysis. 
Since at this point only solutes of low partition ratio remained in the tram 
only aliquots of the lower were examined. The weight/ml uas 
plotted against tube numbers or transfer numbers, as shown in Figuje 
It is necessary to analyze only enough tubes to give the cuive leha \, 
every 5th, 10th or 20th tube, depending on the width ot a band. 

_Xhis section was reviewed by J. Delafield Gregory. 


1 Ahrens, E. H., Jr., and Craig, L. C.; Distribution studies; Separation of the 
.. for cou.,te«™f «rfbuUo., a.„. 

3. . 

merit of equilibrium, J. Biol. Chem. D4: 20.), 1J48. 


4. Barry, G. T.; Sato, Y., and Craig, L. C.: Distribution studies: XIII. Sep¬ 

aration and estimation of higher normal fatty acids, J. Biol. Chem. 188: 
299, 1951. 

5. Battersby, A., and Craig, L. C.: To be published. 

6. Bush, M. T., and Densen, P.: Systematic multiple fractional extraction 

procedures. Anal. Chem. 20:121, 1948. 

7. Cornish, R. E.; Archibald, R. C.; Murphy, E. A., and Evans, H. M.: 

Purification of vitamins: Fractional distribution between immiscible 
solvents, Indust. & Engin. Chem. (Indust. Ed.) 26: 397, 1934. 

8. Craig, L. C., and Craig, D.: In Weissberger, A. (ed.): Technique of Organic 

Chemistry (New York: Interscience Publishers, Inc., 1950), Vol. Ill, pp. 

9. Craig, L. C.; Golumbic, C., Mighton, H., and Titus, E. J.: Distribution 

studies: III. The use of buffers in counter-current distribution, J. Biol. 
Chem. 161:321, 1945. 

10. Craig, L. C.; Hausmann, W., Ahrens, E. H., Jr., and Harfenist, E. J.: Au- 
' tomatic countercurrent distribution equipment. Anal. Chem. 23: 1236, 

11. Craig, L. C.; Mighton, H.; Titus, E., and Golumbic, E.: Distribution 

studies: Purity of synthetic antimalarials. Anal. Chem. 20: 134, 1948. 

12. Craig, L. C., and Post, O.: Apparatus for countercurrent distribution. 

Anal. Chem. 21: 500, 1949. 

13. Fried, J.; White, H. L., and Wintersteiner, O.: Germidine and germitrine, 

two new ester alkaloids from Veratrum viride, J. Am. Chem. Soc. 71: 

14. Golumbic, C., and Orchin, M.: Partition studies: Partition coefficients and 

ionization constants of methvl substituted pyridines and quinolines, J. 
Am. Chem. Soc. 72:4145, 1950. 

15. Golumbic, C.; Woolfolk, E. O.; Friedel, R. A., and Orchin, M.: Partition 

studies on phenols: I\. Isolation of indanols from coal hjMrogenation 
oils, J. Am. Chem. Soc. 72: 1939,1950. 

16. Grubhofer, N.: Neue apparatur zur fractionierten gegenstromverteilung 

zwischen zwei flussigkeiten, Chem.-Ing.-Tech. 22: 209, 1950. 

17. Harfenist, E. J., and Craig, L. C.: Countercurrent distribution of insulin, 

J. Am. Chem. Soc. 73:877, 1951. 

18. Karlson, V. P., and Hecker, E.: Zur mathematischen behandlung der geg- 

enstromverteilung, Ztschr. Naturwiss. 5b: 237, 1950. 

19. Lieberman, S. V.: Simplified calculations of theoretical distribution and 

partition coefficients for countercurrent e.xtractions, J. Biol. Chem 173- 

20. Lochte, H. L., and Meyer, H. W. H.: Medium-sized laboratory apparatus 
for distribution analysis of mi.xtures. Anal. Chem. 22:1064, 1950. 
ichols, P. L.: Useful relations for countercurrent distribution computa¬ 
tions, Anal. Chem. 22: 915, 1950. 

22. 0’Keeffi,^Y E., el a/.: The inhomogeneity of heparin, J. Am. Chem. Soc. 71: 

23. O’Keefe, A. E.; Dolfiver, M. A., and Stiller, E. T.: Separation of the streo- 

tomycins, J. Am. Chem. Soc. 71: 2452, 1949 



26. Swart, E. A.; Hutchison, D., and Waksman, S. A.: Neomycin, recovery 

and purification. Arch. Biochem. 24:92, 1949. 

27. Titus, E., and Fried, J.: Countercurrent distribution studies on strepto¬ 

mycin: The tautomerism of streptomycin, J. Biol. Chem. 174; 57, 1948. 

28. Titus, E., et al.: Distribution studies: IX. Application to metabolic studies 

of 4-aniinoquinoline antimalarials, J. Org. Chem. 13: 39, 1948. 

29. Tschesche, R., and Konige, H. B.: Apparatur zur gegenstromverteilung fiir 

preparative zwecke, Chem.-Ing.-Tech. 22: 214, 1950. 

30. Wevgand, F.: Einfache apparatur zur gegenstromverteilung, Chem.-Ing.- 


31. Williamson, B., and Craig, L. C.; Distribution studies: V. Calculation of 

theoretical curves, J. Biol. Chem. 168: 687, 1947. 


WILLIAM STEPKA, University of Pennsylvania 

Two MAIN features distinguish paper chromatography from classic 
methods of analysis. (1) Remarkable selectivity makes it ideally 
suited for analysis of multicomponent mixtures. With proper pre¬ 
cautions unambiguous results freipiently can be obtained eyen if 
the mixtures contain such closely related substances as homologues. 
(2) The possibility of performing a qualitatiye analysis on micro 
amounts of sample permits, for example, determination of amino 
acid residues from as little as 300 /xg of protein (33, 94, 138). This 
possibility is often advantageous in other applications, notably in 
studies of intermediary metabolism with tracers. Here a radio¬ 
autograph of a chromatogram not only shows the minimum num¬ 
ber of radioactive compounds involved but also indicates the 
scope of the problem confronting the investigator. Frequently 
the probable nature and identity of the spots can be established 
with little extra effort. Such chromatographic information can be 
of great value in the task of ultimate isolation and characterization 
In many applications, such as the study of soluble fractions from 
tissues, paper chromatography allows analysis without intro¬ 
duction 0 reagents stronger than those used for development 

f or complex mixtures likely to contain labile components, the 
virtue of this is evident. 

Introduction of filter paper into partition chromatography by 
Consden, Gordon and Martin (33) followed earlier work of Martin 
and Synge (9/), who conceived the idea of separating components 
of a mixture by hquid-hquid partition in a countercurrent machine 
and in common with previous workers foresaw its geTei-rapDr 

;tcSef fed t It 

63, 94, 98) and othl-s (150)'fteTrnTy Wah "Thfil'T 


«hip at'the of an AEG Predoctoral Fellow- 




of the mixture on a paper support." Capillarity or capillarity aided 
by gravity causes movement in the predetermined direction. 

A single passage of the solvent through the spot produces a 1-way 
or “strip” chromatogram. When, however, the origin contains 
numerous substances a single excursion will rarely result in com¬ 
plete separation of the components. This ideal can be more nearly 
realized with the 2-way chromatogram. Here, the mixture, applied 
near 1 corner of a rectangular sheet, is developed initially as a 1- 
way chromatogram. Then by turning the sheet through 90°, the 
sejjarations achieved in the first run become a series of parallel 
“origins” for development with another solvent. If properly chosen 
the second solvent may resolve the overlaps from the secondary 
origins. Such double development produces the 2-way chromato- 
g)'am with its characteristic pattern of spots. 

Consden et al. (33) regarded the separation of solutes as a func¬ 
tion of their partition coefficients between 2 liquid phases. Accord¬ 
ingly, the paper is represented as an inert support which, in effect, 
immobilizes 1 member of the solvent pair (water), holding it sta¬ 
tionary while the organic member slowly flows past in a thin liquid 


^^dth this concept the process responsible for separation can be 
imagined as follows: At any instant solute molecules at a given 
position on the paper become distributed between the stationary 
and mobile phases in accordance with their partition ratio in the 
2 phases. Since the organic phase is moving and a continuous supply 
is available from the source, solute which passed into this phase at 
location A will be carried along to another but proximate position 
B. At B, however, no solute molecules are initially present in the 
untlerlying water phase, hence solute will pass from the organic 
phase into the water phase until the respective concentrations pre¬ 
vailing at B are again in accord with the distribution law. thus, 
bv continuous adjustments of distribution equilibria at successive 
sites, a given solute is transported from the area of origin to some 
other area aloiiR the path of tlie advancing phase 1 he distance 
througl) Avhich any solute will move is governed by the latio ot iL 
instantaneous concentrations in the 2 phases. Migration is avore 
bv an increase of the fraction in the mobile 

attributed largely to 2 factors: (1) P^P-^,f. 

numerous sites at which adjustments of eciuilibiia can occui 
(2) the rapid establishment ot each equilibrium, since only nunute 
^tanc^ need be involved in the transfer of 

The ratio of the distance through which a compound travel 
the ^tate through which the mobile phase has moved, measured 



. from the point of application or origin, has been called the Rj value 
(33). Under given conditions of solute concentration, solvent pair, 
paper, pH and temperature, this value remains fairly constant for 
any one compound (33). Consequently it is possible to prepare 
chromatographic maps (33, 46, 110) showing the expected posi¬ 
tion of a given compound relative to other compounds (Fig. 1). 
The effects of variables are considered later; of interest here is the 
bearing Rf values have on the interpretation of the mechanism 
involved in the separation. 

Consden et al. (33) derived partition coefficients for certain 
amino acids from their observed Rf values and compared them 
with those determined classically by England and Cohn (Table 
1 in (33)). Close agreement led them to rule out adsorption phe¬ 
nomena and to explain movement and separation solely on princi- 

TABLE 1.—Comparison of Measured and Calculated Rf Values in 
Butanol-Propionic Acid-Water Solvent* 










0 78 



0 64 



0 47 



0 4.3 



0 40 



0 38 



0 3fi 




* Taken from Bassham (7); cf. also Benson et al. (10). 


0 20 

pies of liquid-liquid partition. Synge (99, 143) working with amino 
hrJtpretoiom gramicidin similarly supported this 

rX.5/- ^ “'rasured distribution coefficients of several 

n^dioactive organic acids, glucose and alanine in a butanol-pro- 
Pionic acid-water system. R, values were determined froVchro- 
matograms developed m the .same solvent. Using the value deter 

terms^^krtheT”®.^" ® bassham evaluated the unknown 

tribution coeErxil to dis- 

Rf = 

excursion of band 

Al + aAs excuTiEU^n^^^b]]^-^^ 

where d, = fraction of cross-sectional area occupied by mobile 



phase; = fraction of cross-sectional area occupied by stationary 
water phase, and a = distribution coefficient = (cone, of solute in 
stationary phase)/(conc. of solute in mobile phase). Multiplying 
both numerator and denominator by l/A^, gives 

1 + oiAs/Al 

The term is constant for a given set of conditions (paper, 

solvent, temperature); a is a characteristic of the solute and 
determines the value of Rf, i.e., position on the paper. 

For alanine, the measured value of a in the n-butanol-propionic 
acid-water system was 3.05; Rf as measured from the chromato¬ 
grams was 0.33. Substitution into the equation permitted evalua¬ 
tion of the constant. As/A l = 0.55. R; values for other solutes in 
the same system may now be predicted from their distribution 
coefficients according to the equation 

"" 1 + 0.55« 

Table 1 affords a comparison of Rf values calculated from the 
eciuation by substituting independently determined values of a 
with those measured directly from the chromatograms. The data 
offer evidence in support of the conclusion (33) that the paper 
functions primarilv as an inert support and that a is the piincipal 
parameter which determines the Rf value. Bate-Smith and Westall 
(9), from a study of the chromatographic behavior of phenolic sub¬ 
stances and derivatives, similarly support this interpretation. 

Not all published data agree, however, with an explanaUon ot 
function based entirely on liquid-liquid partition. A number ot 
deviations from predictions according to this theory were noted 
t,V Moore and Stein (see Table 1, p. 272, in (102)) working ^ 
amino acids on starch columns. Taking data from ( onaden ct n . 
(3.1), thev cross-compared K/ values on paper, band rate.s on stare i 

and directly determined partition “o 

them to question the applicability of the ‘he®.y to 

either starch or paper. Additional experience confirmed their 

'^°By Vrimple experiment Moore and Stein (Fig. (i in (102)) 
effectively Temon.itrated that adsorption does function on a 
starch column. They achieved complete separation "f 
and glvcine using onlv water as developing solvent. Likemse 
leucine phenylalanine and tryptophane were separated with eithei 
water or 0 IN Ht’l. With these simple solvents the band ra es. 
:ons'Vemble aLleration. This makes it .lifficult to assess the data 


for the degree of adsorption with the usual binary or ternary sol¬ 

Arden and co-workers (2), also using water miscible solvents for 
the separation of inorganic ions, agree that the liquid-liquid concept 
is inadequate for a complete explanation of the process on paper 
supports. Lugg and Overell (89) were obliged to conclude from their 
observations on organic acids that adsorption is considerable in the 
absence of suppressing agents. Finally, by showing that amino 
acids can be recovered from areas of the paper over which they 
passed during development, Jones (77) showed conclusively that 
adsorption does occur. Other workers who discussed the problem 
(12, 40, 64, 83) considered the process more involved than simple 
liquid-liquid partition. Martin (95a) has reviewed some aspects 
of the problem. 

This treatment is not primarily concerned with the present 
highly controversial subject of the theory of paper chroma¬ 
tography. It would appear better to leave the question open and 
consider that the relative rates of movement of the spots are 
governed as postulated by Consden et al. (33) except that the 
partition ratio need not be restricted to that of a liquid-liquid 
system. What matters is the over-all partition ratio in the presence 
of the adsorbent or supporting agent. A solid support is always 
present, it niay, or may not, play a role in determining the parti¬ 
tion ratio. Similarly an aqueous phase, which alone would be im¬ 
miscible in the mobile phase, may be present. Yet its effect in 
determining a and the Rf value could be secondary to that of 
adsorption. Solid-liquid and liquid-liquid partition ratios often 
depend on similar properties of the molecule. Consequently correla¬ 
tion of 7?^ values with liquid-liquid partition ratios cannot be en¬ 
tirely conclusive for ruling out adsorption effects. 

The broader view is better from the practical standpoint since 

vents Tl T restricted to water-immiscible sol- 

ents (11, 12, (it). In certain applications water-miscible solvents 
as propanol and ethanol, or even water itself, ma,v be superior 

Whatever the principles involved, the utility of paper chroma 
164) or to 111 vitro enzymatic studies (14 75 ’iqo ’ 1 h ’ 

«aXrrev*;:::^ tbe^r 

extension to tracer technique (3, 7, 10, it, 



142, 145, 146, 148, 149), use of the method in metabolic studies is 
bound to increase. 

Used with tracers, it shows great promise since the combination 
is capable of revealing intermediates at concentrations which by 
conventional means would have a low probability of detection. 
Because duration of exposure can be regulated in accordance with 
the intensity of radiation, x-ray film is both the most sensitive and 
the most convenient means of detecting spots of weak activity. 
Spots containing only a few counts above background of C'^ 
radiation expose the film weakly but detectably in about 3 weeks. 
Finding these by scanning with counting devices would be in¬ 
expedient. After exposure, the processed film provides a key to the 
location of the radioactive compounds, which may then be counted 
in situ or excised and eluted for further study. 

Paper chromatography may be used to provide evidence of the 
absence of a stable compound from biologic materials. Thus Cons- 
den et al. (38) disproved the report that nor-leucine occurs in spinal 
cord. Likewise, Joslyn and Stepka (78), by setting an upper limit 
for the concentration of dopa in extracts, adduced evidence against 
its participation in the browning of fruit tissues. Information of this 
sort may either save efforts to isolate a nonoccurring compound 

or indicate an alternate approach. 

It should not be interred from this that papei chiomatography 
by itself is capable of detecting every component in any prepara¬ 
tion. The quantity of any 1 compound on a chromatogram is 
governed by the amount of starting material that may be applied 
to the origin without overloading. A component may therefore be 
present below detectable limits unless fractionated and concen¬ 
trated before application. . - 

Already almost all the groups of biologically important com- 
pounds have been investigated by the paper technique. A partial 
list includes: adrenalines ( 72 ); alkaloids ( 105 ); anthocyamns ( , ), 
inorganic ions (2, 23, 45. 118); flavines and their nucleotides (41 , 
angiotonin (hypertensin) (49); keto acids (2i , 82 91); organic acids 

(10 19,20,79,89, 1,34) ; phosphateesters (10,31,04), penicillins (6 , 

157); porphyrins (126); purines, 

(50, 68 70, 92, 93, 131, ISI^ISS); steroicl^s (24, 

streptomycin (09); sugars (50,56, 6o 61 , 13, / , ^ 

may be extended by consulting reviews (13, 30, 30a, 32, 39, 90, 

®^Ap^Lion to the study of amino acids, either free or from pro¬ 
tein and peptide hydroly-sates ^’o^’ud ’ 14^! 

45-49, 63, 76-78, 94, 99, 116, 119. 120, 124, 128, 1.38 140, 144, , 





retie fractionation (34); resolution of the fractions on paper; elution 
of the individual peptide; hydrolysis of an aliquot of the eluate 
followed by chromatography to determine component amino 
acids. End-groups are determined by deaminating another aliquot 
of the eluted peptide with nitrous acid, hydrolyzing the deaminated 
peptide, and again chromatographing the hydrolytic products. 
Comparison of the 2 chromatograms identifies the terminal 
residues by their absence from the 2d chromatogram owing to the 
loss of their amino groups. Full details are given in the original 
papers (3(), 37). Dent (45) has described a similar method for end- 
group determination. Sanger (128) used 1 ;2:4-fluorodinitrobenzene 
for the same purpose. Jones (77) and Sanger (in (13)) reviewed the 
subject of peptide analysis. 

Choice of method .—Whether the usual descending (33) or the 
more recent ascending method is chosen depends in part on the 
nature of the problem and the equipment available. Williams and 
Kirby (155), who introduced the ascending method, claimed for it 
the advantage of yielding sharper spots. However, experience 
indicates that the descending method can produce spots of equal 
sharpness. Within limits, better definition is obtained with slower 
rates of solvent flow which the ascending method automatically 
provides. Still, whenever it is necessary to employ the full area of a 
standard filter paper sheet the slight improvement attributable to 
this factor scarcely justifies the extra time required for ascending 
chromatograms (for comparative rates see Table 1 in (155)). 
That adequate resolution of complex mixtures requires full utiliza¬ 
tion of large sheets can be inferred from Figure 1. Other variables 
besides rate of solvent flow contribute to spot shape (33, 46, 77, 
104) ■ hence on considerations of the time factor and possible com¬ 
plexity of the sample, the advantage is with the descending 

wooden bars resting on the suppo 
In this way a single cabinet can 
cylinders do not have to be forr 
space conserved. 




a) Cabinets .—These should be vapor-t%ht to assure a saturated 
atmosphere with respect to the developing solvents (33). Unless 
cleanable, it is best to reserve a separate cabinet for each solvent. 
The cabinets described here are a modification of Dent’s design 
(46). They are built from V 4 in. plywood to the following inside 

— Gloss Rod 





35fnm Gloss Tubing- 



F.o. com,K>„ents of (rougl, assembly; e, components assembled. For 

description, see text. 

dimensions: 27 in. long, 18 in. wide and 26 in. deep. Cleats IV V 

, 4 m., attached to the inside of the 18 in will 3 in Koi n 

serve as rests for the trough bases (Fi^ 2 d a ! "'T' 

strin 1 V 9 i'*i fl v -iu icmforcing wooden 

■n. B.I, 




beted groove along the rim. Trunk hasps or sash fasteners, 2 per 
side and 1 per end, secure the lid. Locks are attached to the rein¬ 
forcing strip; the striker plates to the edges of the lid. Two asym- 
metrically placed guide pins in the reinforcing strip, matching cor¬ 
responding holes in the lid, assure proper fit and orientation of the 

The inside of the cabinet can be heavily impregnated with paraf¬ 
fin; but Formica paneling, found to be resistant to most solvents 
used, makes a superior lining. Paraffin-lined cabinets may require 
“aging” in the presence of the solvent before fully reproducible 
chromatograms are obtained. 

b) Trough assembly (Fig. 2, e ).—The trough base d is easily 
constructed from blocks of wood and 26V 2 in. lengths of 6 mm 
glass rod. The notches in the end plates are spaced to hold the anti¬ 
siphoning rods 6 about Vs in. above the rims of the trough c. Papers 
hanging directly over the rim are prone to siphon. The antisiphon¬ 
ing rods are also made from 26 V 2 in. lengths of 6 mm glass rod. 
Short pieces of rubber tubing near the ends prevent the rods from 
slipping out of their notches during manipulation. The weight a 
should be made from heavier rod or from 6 mm tubing filled with 
lead shot. This assembly, like the self-contained troughs of Steward 
et al. (141; see also, 162), permits loading outside the cabinet. 

For troughs, glass is the one material consistently resistant to 
chemical action of the solvents. Although stainless steel has pio\ed 
satisfactory for some workers (10, t HO, 162), an instance of ex¬ 
cessive phenol oxidation was traced directly to a stainless steel 
trough (133). Whether a given type will be satisfactory for phenol 
solvents seems to depend on its composition; none are completely 
resistant to HCl. This point merits consideration in the choice of 

equipment if use of HCl is anticipated. 

Glass troughs (Fig. 2, c) can be made by sawing through a 
length of 35 mm Pyrex tubing with a diamond saw. Turning up the 
ends in a flame and grinding the sharp rims completes the trough. 
As an alternative, lengths of tubing may be split by the hot wire 
method, or the method of Longenecker ( 88 ) may be used. 

c) Paver .—Stock filter jiaper should be kept scrupulously clean 
and protected from ammonia vapors. Adsorbed ammonia, m t e 
absence of catalvst poisons, accelerates the oxidation of phen 
during the run. This is undesirable because the oxidation products 
if excessive, interfere with even flow of the second solvent. It is 

best to handle the paper only by the edges. ^ nrpfcrred 

Because the fibers of most papers possess a degree of pi efei re 

orieiRation which influences solvent flow, R, and spot shape ( 22 , 

t Polished stainless .steel, Type no. 316. 



77, 104), chromatograms for comparative purposes should all be 
run with the texture oriented in the same direction. In practice this 
is accomplished by observing a standard routine in removing sheets 
from the stock. 

Sheets of Whatman no. 1 or no. 4, I 8 V 4 X 22 V 2 in., are most 
commonly used. Other papers were considered to have advantages 
by some workers ( 22 , 28). Kowkabany and Cassidy (83) examined 
75 papers with reference to suitability for chromatography and 
made recommendations based on their tests. Jones (77) stated that 
on Whatman no. 7 amino acids form round spots. Judging from the 
experience of Muller and Clegg (104), one might expect that on this 
paper other compounds would show a similar improvement in spot 

The effects of metals in the paper were discussed by Consden et al. 
(33, 35), who attributed a dark phenol front and “bearded” amino 
acid spots to the presence of copper. Both disturbances were elimi¬ 
nated when catalyst poisons such as H 2 S or HCN were admitted into 
the cabinet, or if cupron were added to the phenol solvent. Hanes 
and Isheiwood (64) also experienced difficulty with impurities in 
the paper. Besides copper they implicated lead and other cations in 
whose piesence phosphate esters either failed to move or streaked. 
Washing with dilute acids corrected the difficulty. Hanes and Isher- 

wood (64) therefore described an apparatus for washing paper in 

M hether preventive measures of this sort will be necessary can 
be determined with a few trial runs without them. For amino acids 
and sugars It will be found that most batches of Whatman no 1 
and ot similar papers can be used as supplied. 

the many solvents described, water-saturated 
fster I K (’• ■'i'"® (33, 46, 48) and n-butanol-acet,c acid- 

bin f 109) have proved most versatile. No com¬ 

bination of solvent systems in use will separate completely all the 
ammo acids or sugars on a single chromatogram. This is probably 
true for other groups of nearly equal complexity P™Oably 

suga'JrTL''solf " “eeful for the amino acids and 

sugars the solution should be colorless or nearly so- a colored 

h llrPtad: ‘•y ‘<^P°-ting oxidation product on““he ditlStn tmd r 

quent preparation of the solvent Wht neerd“the3 “’’r 
solution is complete. On 



solvent is ready for use. Prepared in these proportions the phenol is 
slightly undersaturated and no aqueous phase should separate at 
temperatures above 20 C. 

Consden et al. (33) described water-saturated s-collidine (2:4:6- 
trimethylpyridine) as a satisfactory solvent for amino acids. How¬ 
ever, Dent (46) with Stepka and Steward (48) w^ere unable to pre¬ 
pare a satisfactory solvent for amino acids with refined s-collidine 
from American sources. They obtained good results, duplicating 
those with s-collidine from British sources, by mixing 1 vol of 2:4- 
lutidine with I vol of 2:4:6-trimethylpyridine and equilibrating 
the resulting solution with 2 vol of w'ater (48). The top was 
used. Dent (46) refers to this mixture as “collidine.” 

With the ternary system an occasional run was lost as a result of 
phase separation due to temperature fluctuation. To circumvent 
this defect a binary system containing 2:4-lutidine t and water was 
devised (78). A satisfactory combination, closely reproducing the 
amino acid pattern obtained with “collidine” (46), was obtained 
by mixing equal volumes of 2:4-lutidine and water. It is undersatu¬ 
rated with respect to water at temperatures below 25 C. Reducing 
the proportion of water raises the separation temperature. 

The butanol-acetic acid-water solvent of Partridge (110), ef¬ 
fective for the separation of sugars, was tested for the amino acids 
(136). Although the Rf values were low, the results suggested that 
increase of the water content might improve the separations. Ac¬ 
cordingly, combinations from several points along the critical line 
of a phase diagram for the components were tested with 26 amino 
acids. The best pattern was obtained with the following formula 
(136): n-butanol, 100 ml; glacial acetic acid, 22 ml; water, 50 ml. 
Only freshly prepared solvent should be used because the proper¬ 
ties change as esterification proceeds. This, however, does not 
reach disturbing proportions with fresh solvent during the 15 hi 

required for the run. . i . /-n- i ^ • 

The amino acid pattern obtained with this solvent (Fig. ) is 

again similar to the pattern obtained on “collidine chromato¬ 
grams. A fe«- shifts occur, notably threonine, ^ ' 

sine. Neither cystine nor dopa decomposes to y‘cnt 
(46 122) on chromatograms run in basic solvents Cystine may 
occasionally show a “tail” indicative of decomposition, but dopa 
appears to be stable. 

Comment by Werner Hausmann 

I have found (64a) that all the common amino acids can be 
on a rgle Ip witli the ascending technique by 2 -dnnenslonal ch.oina- 

t Koppers Company, Inc., Pittsburgli. 



tography in the following systems. The 1st system is made from 200 ml 
of 2-butanol and 80 ml of 3 per cent aqueous ammonia. The 2d is made 
from 150 ml of 2-butanol, 30 ml of 88 per cent formic acid and 20 ml of 
water. In the ammonium system, which must be run first, the solvent is 
permitted to climb for 48 hr, the paper then dried and the solvent per¬ 
mitted to climb for a second 48 hr. In the 2d direction the solvent is per¬ 
mitted to climb for 24 hr and then repeated for another 24 hr period. 
This procedure gave good separation of the pairs ornithine-lysine, serine- 
glycine, valine-methionine, leucine-isoleucine w^hich were troublesome in 
other systems investigated. 

Although certain of the organic acids chromatograph well on 
phenol/butanol-acetic acid chromatograms (Fig. 1), some tend to 
form elongated spots in the phenol direction. Also, their position is 
subject to greater variation in response to changes in conditions. 
Lugg and Overell (89) attributed this to the different properties of 
the ionic species which may exist. They found that these effects 
could be suppressed by incorporating relatively strong acids in the 
solvents. Thus with acetic acid as a component, the organic acids 
formed sharper spots; formic acid was even more effective. Lugg 
and Overell (89) discussed the factors involved in selecting a suit¬ 
able solvent for the organic acids, gave formulas for preparing sev¬ 
eral solvent mixtures and tabulated the associated values for a 

series of acids. Brown (20) separated some of the carboxylic acids 
as anions. 

The lower fatty a,cids can also be separated on paper chromato¬ 
grams (19). Applied as sodium or potassium salts, the anions can 
be separated in solvents containing ammonium ion. One effective 

vd™n h,T prepared by mixing: ethanol, 1 

Barkerr^Tf*^ *3 In t Kennedy Ld 

cc 94 L In “Kent (1 cc cone. NH.OH in 100 

an too sllwlyTh^v “Its, the anions 

an too slowly. They therefore recommended conversion to the 

ror"aruS'of'''“‘^ first SIn equ'- 

ied a wide rangrof’chains ITh 1 P"? 
solvents of general applicability for su^r sIpiratfonT®'"*' 

.Solvents for separating alkaloids oi? p^pT Weten developed 



and described by Miinier and Macheboeuf (105). They also dis¬ 
cussed the principles for selecting effective solvents for this class of 

References to a variety of solvents for specific applications may 
be found in the reviews already cited (p. 30). 

SPECIAL SOLVENTS. —Complete separations even on 2-way chro¬ 
matograms cannot always be achieved. Solvents capable of resolving 
overlaps are therefore necessary. Most of these have been concerned 
with the amino acids, although the information compiled by Jer- 
myn and Isherwood (74) is useful for the sugars. 

Before an overlap can be resolved it must be located and then 
excised. Any compound which either fluoresces in or absorbs ultra¬ 
violet light can be located by this means. Amino acids can be made 
to fluoresce on paper by prior heating (117). Once located, the ovei- 
lap can then be cut out and eluted according to the techniques de¬ 
scribed by either Dent (45), Martin (94) or Consden et al. (36). 
Rerunning the eluate in a solvent known to separate the compounds 

in question shows which are present. 

Another resolving technique (136) which eliminates elution may 
be used The cut-out is sandwiched to a paper strip between 2 mi¬ 
croscope slides held together by paper clips and the strip developed 
in the usual manner. As long as good contact between cut-out and 
strip is maintained the resulting spots are sharp. For this ff is some¬ 
times necessary to trim the cut-out to the dimensions of the slide. 

Overlaps of the leucines and methionine can be separated by 
running in a system prepared by mixing equal volumes of «-b^nol 
and benzyl alcohol and saturating the solution with water (33) 

(n water-saturated n-butanol (46). Work (164) has found tertimy 
amvl alcohol in the presence of diethylamine capable of separata's 
leudne isoleucine and phenylalanine. When valine and methionine 
werlao they can be separated in the following solvent (136)^ 

'""These techniques are not 

may also be used foi directly to the cut-out in 

ever, to have significance the tests must ue 





The ensuing discussion, directed primarily to 2-way chromatog¬ 
raphy, applies also to the 1-way or “strip’’ method. Although the 
latter may have limited usefulness for certain work, it is valuable 
for exploratory runs to establish optimal aliquots, and for cochroma¬ 
tography in verifications of identity as described in the preceding 

The variety of biologic materials which might be studied allows 
only general remarks about preparation of the sample. Where 
permissible, it is expedient to concentrate the sample so that only 
small aliquots (< 100 ^1) need be applied to the origin. Schedules 
should be designed to keep the quantity of inorganic salts to a mini¬ 
mum. If this is not feasible and they interfere (33), the sample 
may be desalted by the procedure of Consden et al. (36). If only the 
neutral components are of interest, salts may be removed with ex¬ 
change resins (110). 

To prepare the chromatographic sheet the sample is applied 
within a spot of 1 cm radius at 1 corner of the paper. Attempts to 
confine the sample within too small an area may be equivalent to 
overloading and thus result in streaked spots. The origin should be 
close to the solvent source to gain the full advantage of solvent 
travel, but it should not come in contact with the surface of the 
antisiphoning rod. Contact with wet surfaces may cause streaking. 
Similarly, the secondary origins for the 2d run should hang free on 
the far side of the rods. For the apparatus described, the origin may 
be centered at a point 6 cm from the 1st edge and, allowing for 
fanning out ’ during the first run, 7 cm from the 2d edge. 

The sample IS conveniently applied with a micropipet joined to a 
tuberculin syringe for easy control. Aliquots greater than about 20 
/A are be^ applied m successive portions to keep the origin within 
bounds. Operation in a stream of warm air, as from a hair dryer 
speeds preparation. When the sample has been applied the shLt’ 

trough'''''''^ '' ' proximal edge, is prepared for the 

Two sheets can be loaded into each trough. The loading onera 



development to the edge. Adjustment of the solvent level allows 
some control over flow rate. 

For routine 2-way chromatograms the author, like Dent (45, 46), 
prefers to use phenol as the 1st solvent. Although in theory the se¬ 
quence in which solvents are used should make little difference, in 
practice reversing the order does appear to have an effect—usually 
for the worse. Possibly the explanation is to be found in the greater 
effectiveness of phenol in carrying interfering substances in the pa¬ 
per along with the front, thus removing them from their sphere 
of action. 

Addition of reagents such as ammonia, cyanide or coal gas is not 
recommended for routine work. Omission of these addenda pre¬ 
cludes the possibility of spurious spots arising from reaction be¬ 
tween reagent and components of the experimental mi.xture (139). 
Their use is best reserved for special purposes. 

Development is continued until the solvent front reaches the 
edge of the paper, or nearly so. For phenol, this takes 22-27 hr. 
For some purposes it may be desirable to run the front off the paper, 
e.g., to improve separation between poorly separated spots. To 
do this effectively, some means must be provided to prevent a liquid 
bead from accumulating along the bottom of the paper. Otherwise 
solvent “piles up” within the paper and spots move at reduced 
rates. Hanes and Isherwood (64) attached an absorbent pad of fil¬ 
ter paper to the low^er edge of the sheet; other w^orkers (74, 111) 
serrate the edge, providing points from which the solvent could 

Sheets can be removed from the cabinet with least difficulty by 
making use of the antisiphoning rods. Clipping each sheet to its 
rod enables both to be removed simply by lifting on the 
rod, which can further serve as a support for the sheet during t e 

Though drying is essential before the sheets can be rim m the 
2d solvent. Specially designed dryers may be ^ „ 

good fume hood will dry phenol papers in ^8 hr. ‘ 

nnssible effects of heat on the experimental compounds (58, 114), 
drying whh heat between runs is not recommended because d gjs 
rise to considerable discoloration from enhanced oxidation of p 

When the papers appear and feel dry, they may be lun 

the information desired but usually involves detecting the com 
pounds on the chromatogram. 




The methods for revealing compounds on a chromatogram fall 
into 3 categories. Compounds may be detected by their behavior in 
ultraviolet light (fluorescence or absorption), by radioactivity, if 
tagged with tracer, and by the color they form with certain reagents 
when these are sprayed on the sheet. All 3 methods can be used on 
the same chromatogram. Thus in a tracer experiment the chromato¬ 
gram may first be exposed to x-ray film to give a radioauto¬ 
graph. Next, it may be scanned with ultraviolet light to locate 
fluorescing or absorbing areas. Finally, it can be sprayed with a 
reagent to form colored derivatives. 

Amino acids, not normally fluorescent, can be detected by this 
means if the chromatogram has been previously heated (114, 117). 
This method, however, is considered less sensitive than the color 
developed with ninhydrin (77, 139). Owing to their strong absorp¬ 
tion, ultraviolet light is uselul for detecting purines, pyrimidines 
and their derivatives (68, 70, 92, 93, 151). 

In ceitain instances it is useful to have a completely general 
method of detecting organic compounds. One such method, based 
on acid permanganate, was described by Prochazka (123). With a 
0.03-0.05N solution of potassium permanganate, acidified with sul¬ 
furic acid to a concentration of 0.06-0.3N, most of the organic com¬ 
pounds tested could be detected as white spots on a purple to brown 
background. Pacsu et al. (108) used alkaline permanganate for de¬ 
tecting sugars and some sugar derivatives. The iodine reagent of 

capable of detecting a wide range of substances 
ana may piove useful as a general reagent. 

Ninhydrin which forms blue to purple colors with the amino 
recent fofhil ™ “'^idered the most sensitive 

thrfshold,rfoftL sensitivity 

thresholds for the ammo acids and some derivatives. The reaaent 

steam for 5 min brought the colors to their Hnal hrtens!ty““‘ '' 

Contraiy to statements in some texts (p 185 in 11201^ ,>■ K a ■ 

.s not specific for the a-amino acids (46 lOS^ Dent 
extensive list of substances which react tvotallv 1 u® ® 
ammo acids and primary amines Pe ‘"‘eluding non-a- 

The ninhydrin s^y as orig^mUy pmp^dn^’ 
cent of the reagent in butanol h it Pf P^'^d (33) contained 0.1 per 
veniently substituted (139). ’ ethanol may be con- 



For some purposes, such as identification, it is convenient to 
have more or less specific tests. A few for the amino acids follow. 

Pauly’s diazo reagent (115) gives an intense red color with histi¬ 
dine, which is more easily detected than the dusky ninhydrin color 
(46). Phenol, which also reacts, can be removed by washing with 
benzene or acetone. The reagent is prepared by mixing equal vol¬ 
umes of (1) one per cent aqueous sulfanilic acid in IN HCl and (2) 
0.69 per cent aqueous sodium nitrite. The histidine area is sprayed 
lightly with this solution and, when the paper no longer appears 
wet, again with 10 per cent aqueous sodium carbonate. Previous 
treatment with ninhydrin does not interfere, but tyrosine also re¬ 
acts, hence if applied to strips the test should be well controlled 

with markers. 

For detecting the sulfur-containing amino acids, Chargaff et al. 
(28) used an iodine-azide reagent. With it cysteine, cystine, homo¬ 
cystine and methionine are revealed as white spots against a bro^^ n- 
ish background. The first 3 react rapidly, methionine slowly. With 
a modified reagent (71), methionine also reacts rapidly. Contrast 
between background and spot is increased by including starch. 
The background color slowly disappears and the sheet may then 
be sprayed with ninhydrin for the other amino acids. To prepare 
the modified reagent, add 1 g of sodium azide and 10 ml of I-KI 
solution (0.64 per cent I 2 , 3 per cent KI aqueous) to 100 ml of 
boiled 0.5 per cent soluble starch solution. The lodoplatinate rea¬ 
gent of Winegard et al. (156) is more general; it can also detect 
cystathionine, lanthionine and derivatives containing oxidized sul¬ 
fur. Both reagents work best only after the papers have been well 

freed of solvents. 

Arginine and some guanidine derivatives (p. 188 in ( )) may 

be detected on chromatograms (34) by the red color forined wit 
the Sakaguchi reagent (127), but the test is less sensitive than the 
ninhydrin reaction (46). However, since combined aigmine leacts, 
the reagent can be used to detect arginine-containing peptides. 
Horne and Pollard (69) detected streptomycin on 
ine first with 0.5N sodium hydroxide, immediately af^ aid ^Mth 
0.25 per cent a-naphthol and again after 2 mm with 5 per cent so- 

References to others can be found in reviews (67,112). 



Ammoniacal silver nitrate, introduced by Partridge (109, 110) 
as a spray reagent for reducing sugars on paper, is a sensitive gen¬ 
eral reagent for reducing substances. Compounds related to the 
sugars like the ascorbic and uronic acids (110), reductone and 
hydroxytetronic acid (112) also react. Inositol (110) and sugar al¬ 
cohols also form dark spots but do so only slowly. Interfering con¬ 
taminants in some solvents (112) and reducing unknowns from the 
sample impair the utility of this reagent. Norris and Campbell (107) 
listed 15 common metabolites, all reacting with alkaline silver ni¬ 

To detect reducing compounds, the reagent, prepared by mixing 
equal volumes of O.IN silver nitrate solution and 5N aqueous am¬ 
monia, is sprayed on the chromatogram, which is then heated at 
100 C for 5-10 min (109). Black or brown spots mark the positions 
of reducing compounds. Washing in dilute ammonia (p. 410 in (51)) 
or photographic hypo removes excess reagent and prevents dark¬ 
ening of the background on exposure to light. 

Aniline yields colored derivatives with aldoses when heated in 
the presence of acid. Aldopentoses give shades of red, whereas al- 
dohexoses tend to give brownish tones. Partridge adapted the rea¬ 
gent for use on paper by substituting oxalic (112), trichloracetic 
(110) and phthalic (111) acids for strong mineral acids which char 
the paper. The aniline hydrogen phthalate reagent (111) is pre¬ 
pared by adding 0.93 g of aniline and 1.66 g of phthalic acid to 100 
ml of water-saturated butanol. After spraying, the papers are 
heated at 105 C for 5 min. Ketoses may react weakly but aldoses 
react strongly, 1 ^g of the latter being detectable. In the presence of 
low concentrations of phosphoric acid, aniline reacts readily even 
vith the ketoses and disaccharides (21). At low concentrations 
phosphoric acid did not attack the paper. 

Naphthoresorcinol has been used to differentiate ketoses from 
aldoses (110). For this, reagent is prepared by mixing equal 

(6)TnereI\ ’’T' "'‘Phthoresorcinol in ethaLl, 

(0) 2 per cent aqueous solution of trichloracetic acid. Red colors 

a d~ at llrr heated 

y o\en at 110 C for 5-10 mm; aldoses react weakly. By heat- 

ig in a humidified atmosphere the range is extended to pentoses 

and uronic acids, ivhich then form blue colors in lo"l5 im 

aciT(t gr'Tf5rt'’'"'’r7-' '’■y of orthoprs,$im(c 

colorjpr ductsyiifha^^ 

>er P'>enohc leagents ^ can be adapted 

^- amii 

sprayed first 'vith'aqueo.(y'p);rifacid'^^to mid‘tLn\ritTTo 



per cent aqueous sodium carbonate appear as reddish-brown spots 
on a yellow background after heating at 110 C for a few minutes 

Organic acids have been detected with various indicators, but 
for general use Lugg and Overell (89) recommended bromcresol 
green and bromphenol blue. Forty mg of the dye was dissolved in 
100 ml of 95 per cent ethanol and titrated with sodium hydroxide to 
pH 5.5 for BCG and pH 4.8 for BPB. Sprayed with this reagent, 
acidic substances appear as yellow areas on a green or blue back¬ 
ground. For nonvolatile acids chromatographed as the anions. 
Brown (20) used chlorophenol red (0.04 per cent aqueous) adjusted 
to pH 7; bright yellow spots were formed on a mauve back¬ 
ground. Anions of the fatty acids have also been detected on paper 
with indicators (19, 79). 

Alkaline picrate is a sensitive reagent for many substances, par¬ 
ticularly for those with keto groups. Some, such as pyruvic, a-ke- 
toglutaric and levulinic acids, react immediately in the cold. The 
range is greatly extended if the sprayed chromatogram is heated. 
\hvid reddish spots appear against a yellow background. Kostif 
and Rabek (82) reviewed the range and sensitivity of the reaction. 
They prepared the reagent just before use by mixing equal parts 
of 3 per cent sodium hydroxide solution and saturated picric acid 
solution. Keto acids have also been detected as fluorescent semi- 
carbazones after spraying with semicarbazide reagent (91). 

A method for the detection of acyl phosphates after conversion 
to hydroxamic acids has been described (132). The hydroxamic 
acids were detected and estimated by their colored derivatives with 
ferric iron. Other esters could conceivably be brought within the 

.scope of this technique (p. 358 in (51)). 

Bioautography, a method employing biologic means lor detect¬ 
ing compounds on paper chromatograms, is especially ^|seful in 
chromatographic studies of growth factors and antibiotics 1 he 
method is potentially cpiite selective, depending on availability 

and choice of the appropriate organism. The 
steii and Eigen (158) and of Long and Williams (87) fllustiate tlu 
use of this method and provide additional references. Gin and Fra- 
.sad (59), applied the same principles to detect enzymes separated 

on paper chromatograms. 


Casual i,.spection of a paper chro.uatog.ani may use ..1 
lieuristie information, hut since the chi-omatosiam 
ouentlv used for qualitative analysis of complex mixtures, t us pui- 
;riB achited inly rvhen the compou,.ds on the ch, 



ha\'e been correctly identified. Correct identification poses many 
problems. Simple measurement of Rf values does not suffice, not only 
because these are subject to variation from uncontrollable or un¬ 
known causes but also because superposition of spots from comple.x 
mi.xtures is a possibility not to be overlooked. Radioautographs 
from work on the CO 2 assimilation by green plants (26) show that 
still unidentified compounds overlap some of the known com¬ 
pounds plotted on the map of Figure 1. 

These considerations do not vitiate the method; they do empha¬ 
size the need for corroborative tests. A combination of several tests 
will usually be necessary to accumulate enough supporting evidence 
for fiducial significance. Dent (46) and Steward and Thompson 
(139) discussed confirmatory techniques applicable to amino acids. 
In the following discussion a general approach will be adopted. 

The pattern of spots obtained by spraying the paper with an ap¬ 
propriate reagent will, by comparison with a map of standards, 
furnish a basis for tentative identification. Differences in the colors 
formed (122) provide additional clues. Thus for amino acids 
sprayed with ninhydrin, the prolines appear as yellow spots; d-ala- 
nine and aspartic acid, especially after basic solvents or with modi¬ 
fied reagent (162), as blue spots. With e.xperience, other differentiat¬ 
ing characteristics for particular compounds may be recognized. 
If a spot does not correspond to a known position,' a guess to possi¬ 
ble identity can be made on the basis of certain relationships be¬ 
tween structure and chromatographic position. Stewai'd and 
Ihompson (139) discussed the principles for the amino acids and 

related compounds. Similar considerations apply to the suffar se¬ 
nes (110). ^ 

From a tentative identification one may proceed to build up evi- 

hyde (113) may be used for glycine. Even if a reagent is not com¬ 
pletely specihc It may still he used in situations where the interfer- 
mg substance(s) can be ruled out. 

provisionally identified spot may be cochromatographed with 
an authentic specimen of the known. Identical behavior in several 

dencTthar^n™" different properties provides strong evi¬ 

dence that known and unknown are the same. This techniaue he 

toTThen** sensitive if the compound being identified is rldioac- 
tive. Then comparison of the radioautograph with the sumved 
chromatogram provides visual evidence of simihritv in ° 


termine whether a compound is acidic, basic or neutral (46, 139). 
Under acidic conditions the Rf values of weak acids increase, those 
of weak bases decrease; contrary relationships obtain under basic 

When the compound to be identified is an amino acid, especially 
from animal sources, it is necessary to show that the spot is stable 
to acid hydrolysis. The great variety of peptides makes it possible 
for every amino acid position to be covered by a peptide (46). 

Hydrolysis, either chemical or enzymatic, can be applied as a 
confirmatory test of identity. An eluate of the tentatively identified 
compound can be hydrolyzed under the conditions required to split 
the corresponding known and the hydrolysate rechromatographed. 
Finding the expected products is then added evidence for the cor¬ 
rectness of the original identification. Enzymatic methods appear 
promising in this application since with them a degree of specificity 
can be achieved. 

D-amino acid oxidase sprayed on chromatograms of insect blood 
was used by Auclair (4) to demonstrate d-amino acids in the blood. 
The keto acids liberated by the oxidase were detected with 2,4-di- 
iiitrophenyl hydrazine. Jones (76, 77) and Synge (144) also used 
the enzyme to identify d-amino acids on paper chromatograms. 

For radioactive compounds the hydrolytic method can be ex¬ 
tended beyond mere observation of scission products. Rate curves 
can be determined—sometimes on material from only a single spot. 
Comparison with curves obtained from authentic mateiial again 
provides a criterion of identity. Further, the eluate from a ladioac- 
tive spot may be converted to a derivative and, with added known 
carrier, submitted to a series of purification procedures (81). Un¬ 
changed isotopic concentration provides another standaid foi judg¬ 
ing identity. 

Chemical derivatives of the eluate also can be compared chioma- 
tographically wJth like derivatives of an authentic specimen in 
tests of identity. Dent’s procedure (46) of converting methionine to 
its sulfone and cystine to cysteic acid illustrates the method. By 
demonstrating chromatographic identity of the ureides prepare 
from eluates of an unknown in potato extracts and from 7 -amino- 
butyric acid. Steward ct al (140) obtained supportuig evidence for 
the identity of the 2 compounds. Hirst and Jones ( 6 /) cite unpu )- 
lished work of Hough et al. in which the latter identified sugars ly 
the chromatographic products found after epimerization. 

Disappearance of a compound followmg some '^em, ai 

treatmUt is still another way of obtaining confirmatory data. 
U^ing this procedure, Norris and Campbell (107) produced evidence 
for 7 he intomediary’role of 2 -ketogluconic acid in glucose oxidation 


by Pseudomonas aeruginosa. Grumpier and Dent (42) used the 
principle to distinguish between a- and non-a-amino acids. 

Physical methods such as x-ray diffraction (29, 43, 125), ultra¬ 
violet (70, 92) and infra-red (125) spectroscopy offer further means 
of testing identity. X-ray diffraction as applied to eluates from 
paper chromatograms is described by Christ et al. (29). 


The operations involved in performing a qualitative analysis on 
paper sheets are summarized here in sequence. The summary is 
intended as a guide to a typical analysis, but the techniques may 
be modified to suit the requirements of the individual investigator. 

I. Preparation of sample: For convenience in subsequent appli¬ 
cation to the origin the sample should be concentrated so that no 
more than 20-100 /xl need be applied to reveal the desired com¬ 

Extracts of tissues may be concentrated directly, but hydroly¬ 
sates usually must also be freed from excess inorganic catalyst. If 
this is a volatile acid like HCl, the excess may be removed by re¬ 
peated evaporation to dryness over solid NaOH in an evacuated 
desiccator. When Ba(OH )2 has been used to mediate an alkaline 
hydiolysis, the barium may be removed as the carbonate or sulfate. 
If an enzyme has been used to catalyze the hydrolysis, it usually 
need not be removed from the hydrolysate before concentration 
unless the amount added is great enough to interfere with chroma¬ 
tographic separation. In the latter event the enzyme protein may 

e removed after heat coagulation or precipitation with organic 
solvents. ^ 

Preparations may be concentrated either by removal of solvent 
a reduc;ed pressure and slightly elevated temperature or by lyo- 
p lyhzation Ihe concentrates or residues should be made up to 
.ome definite volume with an appropriate solvent. All manipula¬ 
tions should be conducted with due regard for the stabilitv or 
c lemi^l reactivity of the compounds under investigation 
bp « of optimal aliquot: For this the sample may 

quot to be taken for 2-way chromatography 

The application may be snoodpH if f^^ter paper sheet. 




2 ) Sheets are loaded into the trough assembly, transferred to the 
cabinet and irrigated with the 1st solvent (phenol-water). The sol¬ 
vent travels to the bottom edge of the sheet in 22-27 hr. 

3) The sheets are removed from the phenol cabinet and dried in 
a current of air. Time required for adequate drying depends on the 
rate of air flow, temperature, humidity, etc.; 6-12 hr is usually re¬ 
quired. The phenol may also be removed by washing the sheets in 
ether (58). Caution! 

4) The dry sheets are irrigated with the 2d solvent (n-butanol- 
acetic acid-water) in a direction at !)0° to that of the 1st run (15-20 
hr required). 

5 ) The sheets are removed and again dried in a current of air for 
4-6 hr. 

IV. Detection; 1) Radioautography. If the sample contained 
radioactive material the dry chromatogram is placed in contact 
with x-ray film. Depending on the amount of radioactivity present 
and on the minimal activity per spot which it is desired to detect, 
exposure may last from a few hours to 3 weeks. Proper orientation 
of the processed film with respect to the chromatogram is facili¬ 
tated by asymmetrical marking of the chromatogram Avith radioac¬ 
tive ink (10) at 2 points. The radioactive areas can then be accu¬ 
rately located on the chromatop-am simply by superposing the ink 

marks with their image on the film. 

Reagents sprayed onto the chromatogram may show corre¬ 
spondence between radioactive areas and specific compounds as 
revealed by their colored derivatives. Or radioactive areas may be 
excised, eluted and the eluates studied by chemical or enzymatic 

It is possible to obtain quantitative data on the radioactive areas 
bv direct radioassay in situ or, preferably, by measuring the radio¬ 
activity in the eluates after plating the latter according to standard 

radioassay procedures. .u r m on 

2) Ultraviolet detection. When viewed under the light of an 

ultraviolet lamp a chromatogram may fluorescent spots oi 

ouenching areas. The former are seen as bright spots against a 
feebly fluorescent background, whereas the latter (e.g., purines ant 
pyrimidines) appear as dark spots against the brighter background. 
■^Te amtno acids do not normally fluoresce but can be induced to 

do so by heating the chromatogram at elevated tempei a 111 

Colored derivatives. Duplicate chromatograms may each be 
sp?ay“h ryroup specific reagent; e.g., ninhydrin for ammo 

:^[dTfl.dicator fprijs for acids, and P'l-o^-Sln'^uSm 
Individual members of each senes viU, if 



concentration, be revealed by the characteristic reaction. 

Resolution of overlaps: The detection procedures may reveal 
instances of poorly resolved or superposed spots such as valine and 
methionine (Fig. 1). These may be resolved by rechromatograph¬ 
ing of the excised overlap as described under Special Solvents (p. 

VL Identification: Comparison of the position of the spots on 
the chromatogram with a map of standards provides a basis for 
tentative identification. Confirmatory evidence can be built up by 
a combination of tests such as cochromatography, specific 
i-eagents, etc., discussed under Identification (p. 44). 


Although Dent (44) had previously used visual comparison of 
spot strength with standards for approximate quantitative esti¬ 
mates of amino acids, the first systematic attempts at quantitative 
applications appear to be those of Poison el al. (120), who used a 
spot dilution technique’’ as well as a direct colorimetric method. 

In the dilution method the hydrolysate was spotted along the 
paper in a series of dilutions. A similar series from a composite 
standard ot known concentration was also prepared. Both were 
chromatographed simultaneously. After development of color with 
ninhydrin the amount of an amino acid in the hydrolysate was de- 
ermined by finding a spot of equal intensity and area for the cor- 

.TphTu standard series. Later Poison (119) 

estimation of amino acids in bacterial hydroly¬ 
sates. J he method suffers from the need for a large number of 

aci,rduTv It -concentration of a given amino 

ataeMordTSesSilr” superposition poses a .serious ob- 

esHmrdmto q-nntitative 

■scrib; I (T 22 8d Tofi> F T PfP®'' '•’■■““ntograms have been de- 
e.™<dien<t Ibnii?"" ‘‘"’nit wider 

to influence ,epr„du'cib,™yT,d cotor" vieWs^SS*^'^ ‘14?^)' 

account of the rem ii-em»nf. f .-Pv g*™" a general 

Fowden fWf Z u n ^ -K'ant.tative chromatography 
amin" Sl/on ^tor 

of interfering fa^ds The^sseSl^r ^^^tomatic study 

-ement of 



development of color with a slightly modified Moore and Stein re¬ 
agent (103). Concentrations were determined spectrophotometri- 
cally. Under the conditions employed, a linear relationship obtained 
between optical density and concentration over the range 1-10 ng 
of a-amino nitrogen. 

Thompson et al. (147) also described a quantitative ninhydrin 
method wherein the color is first developed directly on the paper 
under controlled conditions and then extracted from the spots for 
colorimetric estimation. Their procedure, however, requires special¬ 
ized equipment and appears to have no advantages over the less 
involved technique of Fowden (58) which at the moment appears 
to be the method of choice. 

Woiwod (159, 160) investigated the applicability of the copper 
method (121) for quantitative estimation of amino acids separated 
on paper. For accurate results the comparative appioach had to 
be adopted since the theoretical ratio for a-amino nitrogen to cop¬ 
per was not realized. The method, applicable to quantities of 1—25 
Mg of a-amino nitrogen, has been fully described and its limitations 
discussed (161). Martin and Mittelmann (96), after exploring and 
discarding other possibilities, also turned to the copper complexes 
and determined the copper polarographically. They indicated that 
further work was necessary for final assessment of the method. 

Keston and co-workers (80, 81) described an isotopic method lor 
quantitative assay of amino acids on paper chromatograms. 1 e 
amino acids were converted to the p-iodophenylsulfony deriva¬ 
tives and chromatographed as such. Using I - ® ' 

agent, Keston d al. (81) were able to locate and count the derivative 
on the paper strip. A simple calculation, counts m resolved band 
divided by counts per mole of isotopic reagent, then gave the equiv 
alents of amino acid. The method is general and can be applied to 
any compound tor which suitable isotopic reagents are availab e, 
provided^the derivatives are separable into discrete bands or spots. 
Heie the gjeater resolving power of the 2-way chromatogram has 

obvious advantages over the 1-way strip. geem 

Quantitative applications to paper-resolved ™ 

to have encountered the difficulties ‘p^rt^be- 

■Ac T,y +V.P field progress has been moie lapid, paiuy 

ra^nt^i=^ematerial can be 
quantitative procedures lor the siigais. Flood and 


plied parallel ypots of the sugar solution along a paper sheet. 
After development, a reference strip was remo\'ed and sprayed to 
locate the bands. With this as a guide, areas for each component 
were marked out and excised from the main body of the chromato¬ 
gram. Each cut-out and a paper blank were separately extracted 
with a small volume of water in a modified Soxhlet extractor. 
The sugar in the extract was estimated either with Somogyi’s 
copper reagent or by colorimetric measurement of the formalde¬ 
hyde obtained by periodate oxidation. For consistent results a cor¬ 
rection based on performance of the reagent with a standard sugar 
had to be applied. 

Detailed accounts of improved reagents and procedures were 
given by Flood et al. (56) and by Hirst and Jones (66). The latter 
based their method on the estimation of formic acid formed in the 
reaction between polyglycols containing vicinal hydro.xyls and peri¬ 
odate. Under the conditions adopted, the yield of formic acid was 
about 96 per cent of the theoretical. After destruction of excess peri¬ 
odate with eth 3 dene glycol, the formic acid was titrated with stand¬ 
ard alkali and the concentration ol sugar calculated from the data. 
The determinations are reported to be accurate within 2 per cent. 

Hawthorne (65) used similar procedures except for the reagent. 
In this method the eluted sugars, redissolved in bicarbonate buffer 
(pH 10.6), were oxidized with h^'^poiodite. E.xcess iodine was then 
titrated with standard thiosulfate. Forty /xg of sugar could be esti- 
rnated \vith an error of about 5 per cent. In this procedure precau¬ 
tions must be taken against the loss of iodine through volatiliza¬ 

The purines, pyrimidines and nucleosides, owing to their charac¬ 
teristic absorption in the ultraviolet, can be assayed after elution 
by conventional spectrophotometric techniques. With these sub¬ 
stances, as Hotchkiss observed (70), artefacts brought about in the 
preparation may assume some importance. For the details of pre- 

procedures, the original accounts (70, 131, 152 
153) should be consulted. v , , , 

Methods for quantitative estimation of keto acids (27, 91) and of 
acyl phosphates (132) have also been described. 

In an attempt to correlate some characteristic of the spots with 

undistorted"spols of 

ammo acids and sugars, a linear relationship between length of snnt 


I ^ lound a correspondence between soot area nnH 

the "tn it tlierefore unlikely that 

IV ,11 extension to chromatograms of bio- 


logic preparations. Spots from these are rarely uninfluenced by 
materials from the sample and by variations in conditions. 

The obstacles to absolute cpiantitative work are numerous, and 
although a few have been recognized and controlled, it is unlikely 
that all have yet been brought to account. Different workers have 
emphasized different difficulties. Fowden (58) found that for amino 
acids prolonged contact with phenol, especially at elevated temper¬ 
atures, results in serious losses. He was unable, in contrast to 
WMiwod (161) and Jones (77), to show low recoveries presumably 
due to adsorption along the path. On the other hand, Woiwod 
(161) did not attribute any serious losses to reaction with phenol 
even though he dried his papers for 3 hr at 80 C. Thompson et al. 
(147) do not state the conditions under which their papers were 
dried, nor how long the amino acids remained in contact with phe¬ 
nol. Presumably, therefore, they do not attach great significance to 
these factors, although they do stress others such as temperature, 
humidity, oxidation (see also (103)) and concentration of ninhy- 

Considering the picture as a whole, it would appear best at pres¬ 
ent—especially for amino acids—to adopt a conservative view and 
to regard the differences in emphasis as reminders of the consider¬ 
able problems involved in obtaining truly quantitative results 
from paper chromatograms. This does not imply that the methods 
at hand are without utility or that attempts to improve on them 
are to be discouraged. The point has been stressed to stimulate im¬ 
provements and to encourage the use of supplementary quantita¬ 
tive techniques until experience indicates other^^ ise. 


Conventional paper chromatography is unsuited for separating 
substances having low solubility in water. Such substances travel 
with the front of the organic phase without separjiting. To ovei- 
come this difficulty, fast-moving compounds have been converted 
to derivatives with improved water solubility (165). Since this ap¬ 
proach has limited application, a number of workers ha^e a - 

tempted to devise » system of general utility T‘‘'“ erse 

comnoimds. These efforts led to the development of the levers 
phase chromatogram, in which a stationary nonpolar phase is sub- 

stituted lor the usual acjiieous phase. ^ oqUh. 

Boscott (17) first applied the principle, using a co 



Zaffaroni and co-workers (24, 166) satisfactorily replaced the water 
phase by such relatively polar organic solvents as formamide or 
propylene glycol. They used benzene or toluene as the mobile 
phase. This system, however, was incapable of resolving cholesterol 
from cholestenone (84). With these, satisfactory separations were 
obtained on Quilon^-treated paper, using methanol, ethanol or 80 
per cent ethanol as solvents. Here the stationary phase was consid¬ 
ered as consisting of the nonpolar ends of the stearic acid residues. 

A similar approach was adopted by Kritchevsky and Tiselius 
(85), who used paper coated with silicone. On treated strips they 
were able to separate 6 steroids, including androsterone, isoandros- 
terone and testosterone. The strips were prepared by immersion 
in a 5 per cent solution of Dow Corning Silicone no. 1107 in cyclo¬ 
hexane, blotting off the excess and drying for 1 hr at 110 C. Fol¬ 
lowing application of the samples, the strips were permitted to ad¬ 
sorb the stationary phase (chloroform). For this, they were kept 1 
hr in an air-tight container saturated with vapors of the heavier 
phase separating from the following mixture; 6 vol water, 10 vol 
abs. ethanol, 10 vol reagent grade chloroform. The chromatogram 
was then developed, using as solvent the lighter, more polar phase 
from the same mixture. 


I he technique described here has been shown by its wide appli¬ 
cation to constitute a powerful analytical tool. It is available for 
even wider use m problems of physiology and biochemistry, espe¬ 
cially 111 conjunction with tracer techniques. But, like any method 
paper chromatography is subject to misuse. The chromatogram 
provides separation, not identification. Identification is a function 
o the investigator, who should take pains when dealing with new 
compounds to avoid being misled by wishful thinking. Inspiration is 
a valuable quality and, although it may ultimately lead to the cor- 
rect answer, by itself it does not suffice to identify a compound. 

appreciate the shortcomings of 
the method belore attaching a particular label to a spot and should 
try to minimize any possible misconception by every available 
ross-check, including other fractionation techniques The resolii 

The biologic worker could easily be led to the vipw 

On pother hand, it may be well to state empha'tieal™thaUylm); 
i btearato chromic chloride, E. I. du Pont de Nemoure & Co., Inc. 



the identity of a substance must he iinetiuivocal there is no substi¬ 
tute for actual isolation of a sufficient quantity of the pure sub¬ 
stance for ultimate analysis, study of physical properties, etc. 
Paper chromatography is one of the most useful supplementary 
tools for attaining this final objective. 

Such a conservative view is based on the realization that many 
new compounds still exist in nature whose behavior on paper is as 
yet a matter of conjecture. Artefacts are often encountered even 
with well known compounds. The full power of the method can 
be realized only when cognizance is taken of its limitations as well 
as of its potentialities. 


1. Acher, R.; Fromageot, C., and Jutisz, M.: Separations chromatograph- 

iques d’acides amines et de peptides; V. Les acides amines indiyiduels 
de I’insuline avec une note sur le dosage de la proline, Biochim. biophj'^s. 

acta 5:81-88,1950. . . 

2. Arden, T. V., et al .: A new method for the separation, detection and esti¬ 

mation of inorganic compounds, Nature, London 162; 691—692, 1948. 

3. Aronoff, S., and Vernon, L.; Metabolism of soybean leaves; I. The se¬ 

quence of formation of the soluble carbohydrates during photosynthe¬ 
sis, Arch. Biochem. 28; 424-439, 1950. ^ . » • i • 

4. Auclair, J. L.; “Qualitative Studies on the Metabolism of Ammo Acids in 

” Dnntnrfll thesis. Cornell University, 1948 (55 pp.). 

13. Biochemical Society Symp. no 
T., and Synge, R. I^- -^L (e' 



14. Blackburn, S.: The action of papain on wool keratin, Biochem. J. 47: 443- 


15. Boldingh, J.: Application of partition chioniatography to mixtures insol¬ 

uble in water, Experientia 4: 270-271, 1948. 

16. Borsook, H., et al.: a-Aniinoadipic acid: A product of lysine metabolism, 

J. Biol. Chem. 173: 423-424, 1948. 

17. Boscott, R. J.: Solvent-treated cellulose acetate a.s the stationarj' phase 

in partition chromatography. Nature, London 159: 342, 1947. 

18. Brante, G.: Iodine as a means of development in paper chromatography. 

Nature, London 163: 651-652, 1949. 

19. Brown, F.: Separation of the lower fatty acids as anions by paper chroma¬ 

tography, Biochem. J. 47: 598-600, 1950. 

20. Brown, F.: Application of paper chromatography to the separation of non¬ 

volatile carbo.xylic acids. Nature, London 167: 441, 1951. 

21. Bryson, J. L., and Mitchell, T. J.: Improved spraying reagents for the 

detection of sugars on pajjer chromatograms. Nature, London 167: 864, 


22. Bull, H. B.; Hahn, J. W., and Baptist, V. H.: Filter paper chromatog¬ 

raphy, J. Am. Chem. Soc. 71: 550-553, 1949. 

23. Burstall, F. H., et al.: Inorganic chromatography on cellulose: 11. The 

separation and detection of metals and acid radicals on strips of ab¬ 
sorbent paper, J. Chem. Soc., pp. 516-528, 1950. 

24. Burton, R. B.; Zaffaroni, A., and Keutmann, E. H.: Paper chromatog¬ 

raphy of steroids: 11. Corticosteroids and related compounds, J. Biol 
Chem. 188: 763-771,1951. 

25. Calvin, M., and Benson, A. A.: The path of carbon in photosynthesis: 

H . The identity and sequence of the intermediates in sucrose svnthesis 
Science 109:140-142,1949. “ ’ 

26. Calvin, M., e<a/.: The path of carbon in photosynthesis: X. Carbon diox¬ 

ide assimilation in plants, Symp. Soc. Exper. Biol. no. V, Carbon Dioxide 
tixahon and Photosynthesis (London: Cambridge University Press 
1951), pp. 284-305. ’ 

34. Consden, R.; Gordon, A. H., and Martin, A. J. P.; 

d using paper, Biochem. 
.: lonophoresis in silica 






















jelly: A method for the separation of amino-acids and peptides, Bio- 
chem. J. 40:33-41,1946. 

Consden, R.; Gordon, A. H., and Martin, A. J. P.: The identification of 
amino acids derived from cystine in chemically modified wool, Bio- 
chem. J. 40: 580-582, 1946. 

Consden, R.; Gordon, A. H., and Martin, A. J. P.: The identification of 
lower peptides in complex mixtures, Biochem. J. 41: 590-596,1947. 
C'onsden, R.; Gordon, A. II., and Martin, A. J. P.: Gramicidin-S: The 
sequence of the amino-acid residues, Biochem. J. 41:596-602, 1947. 
Consden, R., el al.\ The non-identity of Thudichum’s “glycoleucine” 
and norleucine, Biochem. J. 39: 251-258, 1945. 

Cook, A. II.: Chromatographic separation of natural compounds. Nature, 
London 164: 300-302, 1949. 

Craig, L. C.: Partition chromatography and countercurrent distribution. 
Anal. Chem. 22: 1346-1352, 1950. 

Crammer, J. L.: Paper chromatogra[)hy of flavine nucleotides. Nature, 
London 161: 349-350, 1948. 

Crumpler, H. R., and Dent, C. E.: Distinctive test for a-amino-acids m 
paper chromatography. Nature, London 164: 441-442, 1949. 

Crumpler, H. R., et ai: /3-Aminoisobutyric acid (a-methyl-zS-alaninek 
A new amino-acid obtained from human urine. Nature, London 167: 

307-308,1951. . , « -a t f 

Dent, C. E.: Detection of amino-acids m urine and other tluids, Lancet 

2: 637, Nov. 2, 1946. . , ^ . 

Dent C. E.: The amino-aciduria in Fancom syndrome: A study making 
extensive use of techniques based on paper partition chromatography, 

Biochem. J. 41: 240-253, 1947. i 

Dent C. !•:.: A study of the behaviour of some sixty ammo-acids and othei 
nin’hvdrin-reacting substances on phenol-collidine filter-paper chromat¬ 
ograms, with notes as to the occurrence of some of them in biologi¬ 
cal fluids, Biochem. J. 43:169—180, 1948. ivr ♦ 1 

Dent C E.: Applications to Study of Amino-Acid and Protem Metab¬ 
olism, in Williams, R. T., and Synge, R. L 'i v'pr^itv 

Symp. no. 3, Partition Chromatography (l^ondon: C.ambndge Lnive . > 

14 C’ E'^StepL^W^^and Steward, F. C.: Detection of the free amino- 
add^"ViL,t c^ll^by partition chromatosraphy, Nature, I-ondon 

Edran“;.'th\'purificatto 'h™icd 

Pvrimid lies, in Williams, R. 1., and nge, n. \ p„n,hridffe 

£c Xmp. no. 3, Partition Chromatography (London. Cambridge 

University Press, 1950), PP-, /-. i . York: Elsevier 
Feigl, Y.-.QvaUlative Analyses by Spot 1 ests (3d ed., xNev\ lo 

Publishing Co., Inc., 1946). Application of filter paper parti- 


R^'i^r’and Fink, K.: Radiocarbon and filter-paper partition 

Fink, R. M.» nnd riuK, i iqqg 

TpvifneTs., and Morrison, G. A.: Quantitative paper 
Msher, K. 15.. l arsoiife, ^ 1Q48. 

chromatography, Natuie, • Quantitative estuiiation 

Hood, A. E.; Hirst, E. L., and Jones, J. R. H 



of mixtures of sugars by the paper chromatogram method, Nature, 
London 160; 86-87, 1947. 

56. Flood, A. E.; Hirst, E. L., and Jones, J. K. N.: Quantitative analysis of 

mixtures of sugars by the method of partition chromatography: I. 
Standardisation of procedure, J. Chem. Soc., pp. 1679-1683, 1948. 

57. Forsyth, W. G. C., and Webley, D. M.: A method for studying the 

carbohydrate metabolism of micro-organisms. Nature, Loiulon 162: 
150-151, 1948. 

58. Fowden, L.: The quantitative recovery and colorimetric estimation of 

amino-acids separated by paper chromatography, Biochem. J. 48: 
327-333, 1951. 

59. Giri, K. V., and Prasad, A. L. N.; Detection of enzymes by the agar- 

plate method and its application to paper chromatograph}'. Nature, 
London 167; 859-860, 1951. 

60. Goodall, R. R., and Levi, A. A.: A microchromatographic method for the 

detection and approximate determination of the different penicillins 
in a mixture. Nature, London 158: 675-676, 1946. 

61. Gordon, A. H.; Partition Chromatography on Paper with Special Refer¬ 

ence to Quantitative Separations, in Chromatographic Analysis, Dis¬ 
cussions of the Faraday Society no. 7, 1949 (London: Gurney and 
Jackson ), pp. 128-134. 

62. Gordon, A. H.; -Martin, A. J. P., and Synge, R. L. M.; Partition chro¬ 

matography in the study of jjrotein constituents, Biochem. J. 37: 79-86 

63. Gordon, A. H.; Martin, .4» J. P., and Synge, R. L. M.: Partition chro¬ 

matography of free amino-acids and peptides, Proc. Biochem. Soc. 
Biochem. J. 37: xii-xiv, 1943. ’ 

64. Hanes C. S., and Isherwood, F. A.: Sei)aration of the pho.sphoric esters 

on the filter {)ai)er chromatogram. Nature, London 164: 1107-1112, 

64a. Hausmann, Un])ublished observations. 

()o. Hawthorne, J. R.: Micro-estimation of suears seoaratefl on thn fil+or 


74. Jermj^n, M. A., and Isherwood, F. A.; Improved separation of sugars on 

the paper partition chromatogram, Biochem. J. 44:402-407,1949. 

75. Jermyn, M. A., and Tomkins, R. G.: The chromatographic examination of 

the products of the action of ))ectinase on pectin, Biochem. J. 47: 

70. Jones, T. S. G.: The chemical nature of “aerosporin”: III. The optical 
configuration of the leucine and threonine comj)onents, Pi’oc. Biochem. 
Soc., Biochem. J. 42: lix, 1948. 

77. Jones, T. S. G.: The Application of Chromatography to Amino-Acids and 

Peptides, in Chromatographic Analysis, Discussions of the Faraday 
Society no. 7, 1949 (London: Gurney and Jackson), pp. 285-290. 

78. Joslyn, M. A., and Stepka, W.: The free amino acids of fruits. Food Res. 

14:459^07, 1949. 

79 Kennedy, E. P., and Barker, H. A.: Paper chromatography of volatile 
acids. Anal. Chem. 23: 1033-1034, 1951. 

80. Keston, A. S.; Udenfriend, S., and Cannan, R. K.: A method for the 
determination of organic compounds in the form of isotopic derivatives: 
I. Estimation of amino acids by the carrier techniciue, J. Am. Chem. 

Soc. 71:249-257, 1949. 

81. Keston, A. S.; Udenfriend, S., and Levy, M.: Paper chromatography 
applied to the isotopic derivative method of analysis, J. Am. Chem. 
Soc. 69:3151,1947. 

82 Kostif, J. V., and Rdbek, V.: The detection of Jaffe-positive substance m 
blood serum by paper partition chromatography. The problem of ap- 
ijarent creatinine, Biochim. biophys. acta 5:210-223,19o0. 

83. Kowkabany, G. N., and Cassidy, H. G.: Investigation of paper strip 

chromatographv. Anal. Chem. 22: 817-819, 1950. m 

84. Kritchevsky, D., and Calvin, M.: Paper chromatography of steroids, 

J Am. Chem. Soc. 72: 4330, 1950. 

85. Kritchevsky, T. H., and Tiselius, A.: Reversed phase partition chio- 

matography of steroids on silicone treated paper, Science 114.299 3 , 

Sfi La'ndui A J , and Avvapara, J.: Use of modified ninliy-drin reagent in 

quantitative determination of amino acids by paper chromatograpln, 

87. Lotra t’”-'WilZs, Bioautographic stud^ on the 

88 . 

rfliiliv Sci0ncG 10/ .23 24^ 1948. 

T ^ M and Overell B T.: “One-” and ‘two-dimensional 

^""plrtition 'chromatographic separations o^f 
sheet ■’t °D V one^and two-d’imensional paper 

2308-2.309 1950 Chromatographic studies on nucleic 

^h; nucSs of five strains of tobacco mosaic virus, li.o- 

chem. J. 46:513-517, 1950 yiierodetermination of purines and 

"w^mWine; h" btologkal materials, J. Biol. Chem. 189 : 597-hOo, 


94. Martin, A. J. P.: Partition chromatography, Ann. New York Acad, Sc- 

49: 249-264, 1948. 

95. Martin, A. J. P.: Partition chromatography, Ann. Rep. Prog. Chem. 

(Chem. Soc., London) 45: 267-283, 1948. 

95a. Martin, A. J. P.: Partition Chromatography, in Luck, J. M. (ed.): 
Annual Reviews of Biochemistry (Stanford, Calif.: Annual Review.s, 
Inc., 1950), Vol. XIX, pp. 517-542. 

96. Martin, A. J. P., and Mittelmann, R.: Quantitative micro-analysis of 

amino-acid mixtures on paper partition chromatograms, Biochem. J. 
43: 353-358, 1948. 

97. Martin, A. J. P., and Synge, R. L. M.: Separation of the higher mono- 

amino-acids by counter-current liquid-liquid extraction: The amino- 
acid composition of wool, Biochem. J. 35: 91-121, 1941. 

98. Martin, A. J. P., and Synge, R. L. M.: A new form of chromatogram 

employing two liquid phases: 1. A theory of chromatography; 2. Ap¬ 
plication to the micro-determination of the higher monoamino-acids in 
proteins, Biochem. J. 35: 1358-1368, 1941. 

99. Martin, A. J. P., and Synge, R. L. M.: Analytical Chemistry of the Pro¬ 

teins, in Anson, M. L., and Edsall, J. T. (ed.): Advances in Protein 
Chemistry (New York: Academic Press, Inc., 1945), Vol. II, pp. 1-83; 
cf. p. 30. 

100. Mason, M., and Berg, C. P.: A chromatographic method for the detec¬ 

tion of tryptophan metabolites, J. Biol. Chem. 188: 783-788, 1951. 

101. McFarren, E. F.: Buffered filter paper chromatography of the amino 

acids, Anal. Chem. 23: 168-174, 1951. 

102. Moore, S., and Stein, W'. H.: Partition chromatography of amino acids 

on starch, Ann. New York Acad. Sc. 49: 265-278, 1948. 

103. Moore, S., and Stein, W. H.: Photometric ninhyclrin method for use in 

101 of amino acids, J. Biol. Chem. 176: 367-388, 1948. 

. iluller, R. H., and Clegg, D. L.: Paper chromatography: Physical and 
geometric factors. Anal. Chem. 23: 403-408, 1951. 

105. Munier, R., and Macheboeuf, M.: Microchromatographie de partage des 
alcaloides et de diverses bases azotees biologiques: II. Utilisation de 
era^nhinn^'T^^ acides: Considerations sur I’importance chromato- 
chim. bToL 32: des alcaloides, Bull. Soc. 

chromatographic estimation of a-amino- 
acids. Nature, London 161: 763, 1948. 

107. Norris F. C., and Campbell, J. R,: The intermediate metabolism of 
Pse^oimnas aenigmom: III. The application of paper chromatog 
laphy to the .dent.fication of gluconic and 2-ketogluco„ic acids”ntef 
mediates m glucose oxid,,tion, Canad. J. Res., Sect. C 27?253 261) 

‘’GeSdL^pUmfalCpTic^^^^^^^^^ I- 

juice, egg white, and foeta'i blood ol°sheeu "rP' h‘™ '“J”'-™'® W'® 

1948. ^ Riochem. J, 42; 238-250, 


111 . 

112 . 








120 . 
121 . 
122 . 












Partridge, S. M.: Aniline hydrogen jihthalate as a spraying reagent for 
chromatography of sugars, Nature, London 164:443,1949. 

Partridge, S. M.: Partition Chromatography and Its Application to 
Carbohydrate Studies, in Williams, II. T., and Synge, R. L. M. (ed.): 
Biochem. Soc. Symp. no. 3, Parlitimi Chromatography (London; 
Cambridge University Press, 1950), pp. 52-61. 

Patton, A. R., and Foreman, E. M.; Glycine l eagent for jjaper chromato¬ 
grams, Science 109; 339, 1949. 

Patton, A. R.; Foreman, E. M., and Wil.son, P. C.: Do amino acids 
fluoresce on papergrams? Sci('nce 110: 593-.594,1949. 

Pauly, H.; Uber die Konstitution des Ilistidins, Ztschr. physiol. Chem. 
42; 508-518, 1904. 

Peterson, D. H., and Reineke, L. M.: The chemistry of circulin: C'hro- 
matographic isolation of the amino acid constituents with powdered 
cellulose, .1. Biol. Chem. 181; 95-108,1949. 

Phillips, D. M. P.; Use of ultra-violet fluorescence in paper chroma¬ 
tography, Nature, London 161: 53, 1948. 

Pollard, F. H.; McOmie, J. F. W., and Stevens, H. M.: The analysis of 
inorganic compounds by jiaper chromatographj'^: III. A scheme for the 
qualitative analysis of an unknown mixture of cations, J. Chem. Soc., 

Iip. 771-774, 1951. , , . 

Poison, A.: Quantitative partition chromatography and the composition 

of E. coli, Biochim. bioph 5 's. acta 2; 575-581, 1948. 

Poison, A.; Mosley, V. M., and Wyckoff, R. W. G.; The quantitative 
chromatography of silk hydrolysate. Science 105: 603-604, 1947. 

Pone C. G., and Stevens, M. F.; The determination of amino-nitrogen 

using a copper method, Biochem. J. 33; 1070-1077, 1939. 

Pratt, J. J., and Auclair, J. L- 

paper partition chromatography, Science 108. 213 214, 1945 . 
Prochazka, Z.: Potassium permanganate as a nonspecific .'‘‘T 

agent in partition chromatography, ChemickeListy, p. 44, 1950 (English 

Proom" h! and Woiaod, A. J.: The examination, l,y partition chroma- 
t^rW, of the nitrogen metabolism of bactena, J. Gen. M.orobml, 

3: 319-32,. 1049o .,.a,ninobutyric acid in extract; 

utiaoJion a'nd .Sfication, I Biol. Chem. 183:45,-458.1950. 


^ Cambridge University Press, 
SaCchi,' S A new color reaction of protein and arginine. J. Biochem., 

Sanger, F.: The terminal peptides of insulin, Biochem J. 4o. o63 5,4, 

G C T A (ed )• The Chemistry of the Amino and Proteins 

Schmidt, C. L. • 1 . rViMiles C Thomas, Publisher, 1945). 

(2d ed.; Springfield, I .. Fnzvmatic formation of pentose phos- 

Scott, D. B. M., and Cohen, S. S EnYm^tic^|^o. 

phatefrom6-phosphog u(ona^ , gaudies on nucleic 



preparations of Clostridium kluyveri: VI. Reactions of acyl phosphates, 
J. Biol. Chem. 184:769-793,1950. 

133. Stadtman, F.: Private communication. 

134. Stark, J. B.; Goodban, A. E., and Owens, H. S.: Paper chromatography 

of organic acids. Anal. Chem. 23: 413-415, 1951. 

135. Stein, W. H., and Moore, S.: Chromatography of amino acids on starch 

columns: Separation of phenylalanine, leucine, isoleucine, methionine, 
tyrosine, and valine, J. Biol. Chem. 176: 337-365, 1948. 

136. Stepka, W.: Unpublished data. 

137. Stepka, W.; Calvin, M., and Benson, A. A.: The path of carbon in photo¬ 

synthesis: II. Amino acids. Science 108: 304, 1948. 

138. Stepka, W., and Takahashi, W. N.: Amide constituents of tobacco 

mosaic virus protein. Science 111: 176-177, 1950. 

139. Steward, F. C., and Thompson, J. F.: The nitrogenous constituents of 

plants with special reference to chromatographic methods, in Arnon, 
D. I., and Machlis, L. (ed.): Annual Revmvs of Plant Physiology 
(Stanford, Calif.: Annual Reviews, Inc., 1950), Vol. I, pp. 233-264. 

140. Steward, F. C.j Thompson, J. F., and Dent, C. E.: "y-Aminobutvric 

acid: A constituent of the potato tuber? Science 110: 439-440, 1949." 

141. Steward, F. C.; Stepka, W., and Thompson, J. F.: Apparatus for parti¬ 

tion chromatography on paper. Science 107:451-452, 1948. 

142. Steward, F. C.; Thompson, J. F., and Miller, F. K.: The amino acids of 

alfalfa as revealed by paper chromatography with special reference to 
compounds labelled with S^^, Plant Physiol. 26:123-135,1951. 

143. Synge, R. L. M.: Analysis of a partial hydrolysate of gramicidin by 
partition chromatography with starch, Biochem. J. 38: 285-294 1944 

^ AA TJTIVT.Ar_ j1. .1 ^ * 

144. Synge, R. L. M.: A further study of hydrolysates of gramicidin, Biochem 


154. Waley, S. G., and Watson, J.; Rearrangement of the amino-acid residues 

in peptides by the action of proteolytic enzymes, Nature, London 167: 
360-361, 1951. 

155. Williams, R. J., and Kirby, H.: Paper chromatography using capillary 

ascent. Science 107: 481-483, 1948. 

156. Winegard, H. M.; Toennies, G., and Block, R. J.; Detection of sulfur- 

containing amino acids on paper chromatograms. Science 108: 506- 
507, 1948. 

157. Winsten, W. A., and Spark, A. H.: Penicillin types produced by P. 

chrysogenum Q-176, Science 106: 192-193, 1947. 

158. Winsten, W. A., and Eigen, E.: Bioautographic studies with use of 

Leuconostoc citvovorum 8081, J. Biol. Chem. 184:155—161, 1950. 

159. Woiwod, A. J.: The micro-estimation of amino-nitrogen and its applica¬ 

tion to paper partition chromatography, Proc. Biochem. Soc., Biochem. 

J. 42: x.xviii, 1948. 

160. Woiwod, A. J.: Micro-estimation of amino-nitrogen and its application 

to paper partition chromatography. Nature, London 161:169, 1948. 

161. Woiwod, A. J.: A method for the estimation of micro amounts of amino 

nitrogen and its application to paper partition chromatography, Bio¬ 
chem. J. 45: 412-417, 1949. 

162. Woiwod, A. J.: A technique for examining large numbers of bacterial 

culture filtrates by partition chromatography, J. Gen. Microbiol. 3: 
312—318 1949. 

163 Wolfson, W. Q.;’Cohn, C., and Devaney, W. A.: An improved apparatus 

and procedure for ascending paper chromatography on large size hlter 
paper sheets. Science 109: 541-543, 1949. 

164 Work E.: Chromatographic investigations of ammo acids from nucro- 

organisms. I. The amino acids of Cornebacterium diphthenae, Biochim. 
bioph vs. acta 3: 400-411, 1949. 

165 Zaffaroni, A.: Burton, R. B., and Keutmann, E. H.: The application of 

paper partition chromatography to steroid analysis: 1. Ketosteroids, J. 

Biol. Chem. 177: 109-116, 1949. , • 

166 Zaffaroni, A.; Burton, R., and Keutmann, E.: Adrenal cortical hormone^s. 

* \nalysis by paper partition chromatography and occurrence in t e 
urine of normal persons. Science 111: 6-8,1950, 


LEWIS G. LONGSWORTH, Rockefeller Institute for Medical Research 

Electrophoresis is the movement of charged particles in solu¬ 
tion under the influence of an electric field. The most widely used 
method for study of this phenomenon is that of moving boundaries. 
In this method the particles, or ions, are not observed directly but 
their presence is inferred from the contribution that they make to 
some property of the solution such as the refractive index. Thus if 
a conducting solution from which the particles are absent is super¬ 
imposed in a vertical channel upon the more dense one in which 
they are present, at the junction or boundary between the 2 solu¬ 
tions the refractive index changes from its value in the 1 solution 
to that in the other. With the aid of optical methods to be described, 
this change of refractivity with height can be observed and photo¬ 
graphed and the boundary thus located. If, now, an electric cur¬ 
rent is passed through the solutions in the channel the particles 
will move, and the resulting shift in the position of the boundary is 
a measure of this movement. In practice, the channel is U-shaped, 
with the solution under observation occupying the lower portion, 
so that if the boundary in 1 side rises on passage of a current, that 
in the other side descends. 

The velocity of a particle when the electric field acting on it is 
1 v/cm is called its mobility, u. In a given medium the mobility 
is a characteristic property of the ion species or particle but 
generally varies with the nature of the solution. This is especially 
true of protein ions where the mobility depends upon both the 
pH and the ionic strength of the solution in which they are dis¬ 
solved. To obtain definite mobilities, therefore, the initial boundary 
IS usually formed between the 2 solutions that result from the 
dialysis of the protein solution against a relatively large volume 
of an appropriate buffer solution. The dialyzable buffer ions are 

en present on both sides of the initial boundary at concentrations 
corresponding to the Donnan equilibrium (10) and provide a con- 

oi'^Wn r -(k move, on passage 


The moving boundary method is not restricted, however to 
mob,hty measuremeirts. In the case of proteins and rellted ma¬ 
terials It may be used for analysis of mixtures such as blood senim. 




This application is possible because the mobilities of the various 
proteins that may be present in a mixture frequently differ from 
each other. Initially each species occupies the same portion of the 
channel, but on passage of the current the zone in which a given 
protein is present moves at a rate proportional to its mobility 
and thus progressively separates from zones occupied by species 
having different mobilities. Thus, in the ideal case, shown diagram- 
matically in Figure 18, the initial boundary in each side of the 
channel splits, on passage of the current, into a number of separate 
boundaries equal to the number of species having different mobili- 
f ies. Below each of these separated boundaries a species is present 
at the same concentration as in the original mixture but absent 
above and is said to disappear across the boundary. Moreover, 
owing to the close proportionality between refractive index and 
concentration, measurement of the difference in this index across 
each boundary affords an electrophoretic analysis of the mixture. 

As will be shown later, other properties of proteins can also be 
determined with the aid of the method, and it can be used for 
preparative purposes. From the foregoing outline of the moving 
boundary method it is clear, however, that the apparatus and 
experimental procedure must provide (a) for the initial foimation 
of a sharp boundary between the protein and buffer solutions (6) 
for the passage of a known quantity of electricity with a 
minimum of disturbance from the heating effect of the current, 
electro-osmosis and the electrode processes whereby the current 
enters and leaves the solution, and (c) for observation and pho¬ 
tography of the boundaries. The manner in which these require- 
ments are met will be evident from the description, in the follow¬ 
ing section I, of a typical experiment with blood seium m a 0 . 
sodium diethylbarbiturate butfer of pH 8.6. In tins and succeedmg 
sections the experimental procedures presented in some detail ar 
Tnes that have been developed, for the most part in this labora¬ 
tory It is well to remind the reader, however, that the fundamental 
ideL underlying the adaptation of the moving boundary method 
orthrstady of proteins are due largely to one mdiv.duah Arne 
»s (52)! and that the method is justly known as the T.sehus 


I. Electrophoresis in the Tiselius Cell 

... ,1 .„77 Tbe Tiselius cell is shown in perspective 

Assembling the “hes of section / that con- 

in Fig 1, a. With the exception of the ubes ot section r r 



sistaiit flux and consists of 3 sections that may be slid ov^er each 
other along the planes AA' and BB'. With these interfaces lubri¬ 
cated with a mixture of 2 parts petroleum jelly and 1 part mineral 
oil and the sections aligned as shown in the figure, a continuous, 
leak-proof, U-shaped channel, C-C, of rectangular cross-section 
runs through the cell. Figure 1, 6, is a top view of the center section. 
As will be shown later (p. 93), the width, 3 mm, of the channel is 
the critical dimension in controlling convective disturbances, the 
depth, 25 mm, largely determines the sensitivity of the optical 
method since this dimension is pai-allel to the light path, Avhereas 

Fig. 1.— The Tiselius electrophoresis cell. 

the resolving power for particles with similar mobilities deDemis 
mainly on the clear height, 86 mm. ‘ooiiicies depends 

ming the cdh-Following insertion of the metal clips e of Figure 
2 betw(*n the horizontal plates of the center section, the assembled 

ti^nW ed hr'ir™'' l»ttom sic 

llightlfaboveVfi' kl 'extending 

and a 2? 00^17 1 of a 20 ml syringe 

and a ^5 cm 17 gauge stainless steel needle The sharnenod ti,. r 

whose shaft carries a pinion engagingthfr 

the edge of the bottom section. Moreove,.; sincl tt'SH .Slr't 



Fig. 2.—-Electrophoresis cell, electrode vessels and support. 



3 A 







Fir,. 3.—Diagrams illustrating tlio initial formation of the boundaries. 

raised the rack can also be brought to liear on the insert c it it is 
necessary to shift the center section to the left. A similar device on 

the left permits shifting either of these 

The rieht side, say, of section II is now filled with the piotem 
solution fnd the left side rinsed and filled with buffer, the leyeb m 
both sides being above .1.1'. With the aid of neoprene sleeves, N 



Figure 2 , the cell is connected to the electrode vessels and these are 
filled with buffer to a level just below their side-arms. This connec¬ 
tion is less likely to leak if the glass surfaces making contact with 
the sleeves are lightly greased with petroleum jelly. Moreover, 1 
end of each sleeve is provided with a neoprene ring on which the 
remainder of the sleeve can be rolled. With a sleeve thus rolled up 
it is slipped over the lower end of the side-arm, which is next swung 
into position above the top section of the cell. The sleeve is then 
allowed to unroll into the position shown in the figure, thus making 
the desired connection. 

The assembly is now placed in a thermostat regulating at 0.5 C, 
a value whose significance is considered on page 95. After allowing 
a few minutes for thermal equilibration the center section is dis¬ 
placed to the right and the bottom section returned to its original 
position (Figure 3, 6 ). The excess protein solution is then rinsed out 
of the top section and both sides of this section, together with the 
attached electrode vessels, filled with buffer to the level IV of 
Figure 2 . Ihe silver-silv^er chloride electrodes E and E' are next 
inserted, care being taken not to trap air bubbles as the ground- 
glass stopper / is seated. To provide chloride ions for the electrode 
process 25 ml of IM sodium chloride is allowed to flow slowly down 
each of the glass-enclosed silver tubes t and t'. As the buffer around 
the electrode E, for example, is displaced, the liquid rises into the 
hollow stopper p and the excess overflows through c. With the aid 
of a rubber cap the tube t is closed while still filled with solution and 
the capillmy rubber tube c, also filled with liquid, is connected, as 
with the syringe d, this having been previously filled 

sh^w,^•^7h''n'^ clamped m the “compensator.” This device, not 
i^houn m the figure, imparts, with the aid of a small synchronous 
motor, a uniform movement at the desired rate to the pLon of the 
syringe. By turning the stopcock m through 90° from the position 

rct'o.frh fhe**® right-hand side of the apparatutl":^" 
connection vith the compensator being retained. It will be noted 
that an is excluded completely from this side, whereas that on the 
left remains open to the atmosphere. ^ 

Forming the boundaries .—Now bv 77 - .l . 

»™«darieB betwtTthe protrif 

right and beUverz/^dw" ^ 

the syringe at the rate of 3.6 ml/hfC2Tmin wn'f h'ff 

level indicated by the schlieren band at N the 

the boundaries have emerged, however, the‘" hne’s’^orth" 



patterns are recorded, that for 1 side of the channel being shown at 
h in Figure 4. This pi’ocediire is described under Schlieren methods 
(p. 71). 

Testing for leakage .—It is also convenient to test for leaks during 
the compensation. This is done by connecting 1 lead from the power 
supply to an electrode and the other to the cell frame, i.e., the bath 
water. On application of a potential difference of 100-200 v f) 
current of as much as a few microamperes usually indicates an 
improperly assembled cell with the possibility of loss of solution 
during the experiment. 

After shifting the boundary the compensator is .stopped and the 

Pjq 4—Negative and tracing of the pattern 

human serum. 

of the rising boundaries of a 

syringe isolated by turning m of Figure 2 through 45°. If not shifted 
too rapidly the slight spreading of the boundaries that occuis dur 
W this process is due mainly to diffusion and for most purposes 
can be neglected. This is not true, however, if a boundary is foi me 
nRiM^ a stopcock as in some cells^ In this 

into the channel but before current is passed. 



iiectiiig the left-hand side of the channel in Figure 2 must be posi¬ 
tive so that the boundary will rise in this side and descend on the 
right. The value of the current i is determined potentiometrically 
from the potential drop E across a known resistance R, connected 
in series with the moving boundary cell, i.e., i = E/R. If the poten¬ 
tial of the power supply is reasonably steady the current drift dur¬ 
ing an experiment amounts to no more than I or 2 per cent. The 
average of several readings of the current, when multiplied by the 
time, affords a value for the number of coulombs passed that is 
considerably more accurate than this. If precise values of the 
mobilities are not required, the current can be measured with the 
aid of a milliammeter. 

Once the electrophoresis is started the investigator is largely 
free to measure the pH and conductance of aliquots of the protein 
and buffer solutions (pp. 87 and 90) to determine mobilities and 
concentrations from previously recorded patterns (p. 80) and to 
prepare solutions for subsequent e.xperiments (p. 87). 

After the boundaries have separated sufficiently to give well 
defined minima in the gradient curves, the current is interrupted 
and the pattern of both sides of the channel recorded. Figure 4, a, 
is the negative of that side in which the boundaries ascended. If 
none of the separated fractions are to be recovered as described 
under Preparative and Countercurrent Electrophoresis (p. 98) 
this concludes the experiment. ’ 

Patterns can also be recorded wffiile the current is flowing. In 
mobility determinations, however, the 2 patterns from which the 
displacement of a boundary is evaluated should be obtained either 
vith the current flowing or with it interrupted. The temperature 


II. Schlieren Methods 

Optical principles.-Mlhough optical methods based on the 
interference of light are being adapted for work wRV» 
boundaries, and are considered in tL next sectioVfn ^ 7fi 
FoucauIt-Toepler schlieren effect introduced by Tis^hus^^L k 

now widely used. The form that the schlieren methoVlTl^l f In 
observation and photoffranhv nf fho r metPod takes for the 

the boundaries is sS ^rlmstic 'il '‘t 

spherical lens callerfhe scWrener wl r'T" '•- " 

the illuminated, horizontal slit S c»i/ !i of 

camera lens 0 is focused on the channel C and fot 
on the screen at G If the fluid ihi ftio u i fo^ms an image of it 
--■ the Huid in the channel ,s homogeneous this 



image will be uniformly illuminated. Suppose, however, that a pro¬ 
tein-buffer boundary is present. Such a boundary consists of a 
region B, of finite thickness, in which the refractive index n varies 
continuously with the height h from its constant value in the body 
of the protein solution to that in the buffer. The gradient, dn/d/i, 
of refractive index thus varies from zero to a maximum and back 
to zero. 

In Figure 5 the variation of the gradient in the boundary is indi¬ 
cated by the density of the shading. Each layer of solution in the 
boundary acts like a prism and deflects the light passing through it. 
With but few exceptions, e.g., aqueous ammonia, the denser solution 
below the boundary has the higher refractive index and the deflec- 

Yig 5 —Schlieren method for study of refractive index gradients between 


tion is downward. For a sufficiently thin layer the gradient may be 
considered as constant and the light traversing this layer forms an 
image of the slit below the normal one at Fo. In ^>8“^ ^ 

displaced images are indicated. That at T„ is forme y le _aym 
in the boundary at h. having the maximum ^ ft 

at T, is the superposition of the 2 images formed by the 2 layers a 
h and /. for which dn/dh has identical values. Actually the dis¬ 
placed slit taages formed by all the layers in the boundary com¬ 
bine to give a rectangular pattern of interference fringes, 

called diaphragm and 

halU^hTiLtan^^^^^^^^^ is placed m the plane of t e 

f" ‘Tr Ltulfy “ifb: «Cd 

IrFwi rapp c at the cLjugate level ^ in the 

bTudlry ^11 bfhitercepted and the dark band in the image of 



the channel will broaden until the edges of the band are at the con¬ 
jugate levels Hi and Hj. This, obviously, can be continued until the 
undeflected rays at Fo have been intercepted and the entire image 
of the channel becomes dark. 

The scanning process .—With a single boundary in the channel 
the series of 14 photographs shown in Figure 6, a, was taken with 
the schlieren diaphragm raised a fixed distance between each expo¬ 
sure. The progressive broadening of the schlieren band is apparent. 
In the schlieren scanning procedure (22) the stepwise recording 
shown in Figure 6, a, is made continuous. The central portion of the 
life-size image of the channel at G, Figure 5, is masked by a ver¬ 
tical slit 0.2 mm wide and the photographic plate moved horizon¬ 
tally in the direction at a constant rate across this slit. Actuated 

by the same mechanism, the diaphragm D is given a steady move¬ 
ment upward. The resulting photographic positive for the boundarv 
of Figure 6, a, is shown m Figure 6, b, and is called the boundarv 
pattern. The schlieren bands of a correspond to narrow vertical 
sections of the complete pattern. For publication it is common 
practice to turn the pattern through 90°, as shown in Figure 6 c 
If more than 1 boundary is present in the channel, each appears as a 
separate peak in the complete pattern (e.g.. Fig. 4 a) 

As noted above, the edges of a schlieren band are conjugate to the 
ves in the boundary at which the gradient deflects the light 
the same distance, AF, below the normal slit image. To a verv 
se appioximation this deflection is given by the relation 

AF = ah dn/dh 

Mrm"in th'rT-T"”'',?' to ‘he optic axis 

center of the channeUnTlI ’ “nr ^ distance from the 

outline of the 

rhelCmtrtectortVe » channelT^ri: 

this plot is which will be designated as . The oMinateTalt 



from equation (1), ah M dn/dh, designated as N, where M is the 
ratio of the rates at which the plate and diaphragm move in the 
scanning process. The vertical line at Hq, Figure 6, c, is due to a 
graduation ruled on the channel window with a diamond pencil 
and is taken as the origin of abscissas. The origin of ordinates is the 
horizontal line at Nq joining those portions of the pattern outline 
that correspond to layers in the channel in which the solution is 
homogeneous. This is the base line of the pattern. In the ideal case 
this line is straight and horizontal. Actually, however, it is often 
distorted by imperfections in the lenses and in the cell and thermo¬ 
stat windows. Allowance for such distortion is made (described on 
p. 80) with the aid of a scanning photograph of the channel when 
it contains a homogeneous solution. Although it is a rather common 
practice, the recording, simultaneously with the boundary pattern, 
of a “base line” with light that has not traversed the cell cannot be 

In addition to the scanning modification of the schlieren method 
for recording the pattern, there is the cylindrical lens procedure of 
Philpot (41) and Svensson (47). Since a clear account of this 
method is given by Pickels (this volume, p. 114) a description of its 
adaptation to electrophoresis will not be necessary here. 

Limitations of schlieren methods .—Although both the scanning 
and cylindrical lens procedures give the complete pattern, they 
should be considered as supplementary to the schlieren band 
method originally used by Tiselius. Only by observing the distor¬ 
tion of the edges of these bands as the diaphragm approaches the 
normal slit image is it possible to detect the mild convection that 
frequently precedes more serious disturbances. In the canning proc¬ 
ess *e image mask screens all but the central part of the channel, 
whereas with the cylindrical lens the recorded value jrad'etit 

is an average over the entire cross-section of the chaimel. Oui pre 
erence is for equipment that is flexible with respect to the optica 
method employed so that the advantages of each may be utilized 
rnlrkvriS unfamiliar proteins or buffers the weak gradients at 
he edger of a boundary are examined at intervals during he 

experiment If the edges of the schlieren bands have horns at the 

wills or a “chimney” at the center these effects can frequen y 

the pattern. In reproducing the patterns oi 



“line” cuts, the effect has been lost. On the original plates, however, 
the successive diffraction maxima and minima in the illuminated 
region parallel to the edge of the pattern, except at the base line 
and top of a peak, are clearly visible. This is shown in Figure 7, 
which is a microphotometer trace through an edge (26). It is essen¬ 
tial to ex{)ose and develop the plate so as to have a moderate den¬ 
sity in the illuminated portions with the diffraction maxima clearly 
evident. On tracing the pattern one then follows a rather low den¬ 
sity, the pencil line even penetrating slightly into the illuminated 
portion of the enlarged image of the plate, since the edge, in terms 



Fig. ^•“*^I*crophotometer trace through edge of a boundary pattern showing 
effects of diffraction at the schlieren diaphragm. 

oi geometrical optics, is at H in Figure 7. As shown in Figure 12 a 
pattern traced m this manner then agrees quantitatively with that 
given by an entirely different optical method. 

>^ources.~The best patterns are obtained with monochro¬ 
matic light, e.g., that given by the G.E. AH 4 mercurv vanor lumn 
and a yellow filter, Wratten no. 22, or a green E 

- and 

due to ab.sorption”tjle1’rglitTonki he of Pattern. If the effect is 

-itted, e.g., red in the cat of h« ^ “'t'otc 

bpennc se.™, every effort should te made to :i:rirth:™f-t!;^,: 



by high speed centrifugation, say, before electrophoresis. This is 
not always practicable, however, in which case the effects of the 
scattering can be reduced by using red, or infra-red, light from an 
incandescent source (55) and a fast, red-sensitive emulsion of wide 
latitude. Although there is some reflection from the glass envelope, 
a useful lamp for this purpose is the Westinghouse 3.5 amp, 10 v 
galvanometer lamp, the linear filament of which is focused directly 
by the schlieren lens. Alternative procedures are to make 2 expo¬ 
sures, one of the correct density for the scattering zone, the other 
for the clear zone, or to use a bar as schlieren diaphragm. 

Schlieren diaphragms .—Although a straight edge served as the 
schlieren diaphragm in obtaining the patterns of this chapter, a 
slit or a bar may also be used for this purpose in either the scanning 
or the cylindrical lens procedure. A slit gives the pattern as a bright 
band on a dark background, whereas in the case of a bar the back¬ 
ground is light and the band dark. The bars and slits in use, how¬ 
ever, are of such width as to act essentially as 2 parallel straight 
edges, each of which produces diffraction at the plate similar to 
that shown in Figure 7. The resulting band is generally so broad 
that one must trace and planimeter both edges. In the patterns of 
reasonably clear solutions, both edges give the same result and the 
additional w^ork is hardly justified. If, as a result of absorption or 
scattering, the results from the 2 edges differ, the mean values 

should be used. , , 

Schlieren optics.—¥ov the pattern to record accurately the magni¬ 
tudes and positions of the gradients in the channel, the optical sys¬ 
tem should be of good quality. In the case of the scWieren lens the 
requirements are similar to those of a telescope objective. Since 
this lens must have a diameter of 4 in. to cover the ^^ehus cell, 
one having more than 2 elements, i.e., an achromat, is haidly prac¬ 
ticable. In the case of an achromat, however, the image qua i y 
deteriorates at apertures much greater than F/9 so th^ a 4-in. 
achromat should have a focal length of 36 in. 

sideration largely determines the dimensions of the electiophoiesis 
equipment xLJif both the schlieren and camera lenses of Fi^re 5 
are used at unit magnification, the over-all length of the system 

"“sevlrefexpedients have been adopted to make the equipm^ 
more compact. One is to use a schlieren lens on °l,*, j 

rel^taf so that the light ^ 



cells hut not the standard one. IVIirrors are also used, not to shorten 
the optical path but to reflect the light into a more compact space. 

Focusing .—The effects of improper focus (27) of the source slit 
and of the channel are illustrated by a typical boundary in Figure 
8. The pattern of a was obtained with both slit and cell correctly 
focused, a white line being drawn through the maximum gradient 
parallel to the reference line in order to demonstrate the S3^mmetry 
of the gradients in the boundary, h was obtained with the cell 
moved 10 cm toward the camera, a shift from the object plane that 
• causes the boundary to appear to be skew. The lower maximum in 
h is due largelj^ to the shortened distance from the cell to the 
.schlieren diaphragm. After returning the cell to the object plane 
and then moving the source slit 10 cm toward the thermostat the 

1 iG. 8. Effect of improper focusing of 

camera [h) and illuminated slit 


pattern of c was recorded. This di.splacement of the undeviated slit 
ba.i^hnT”^ schlieren diaphragm results in a sloping 

Although the slope of the base line affords a sensitive test for the 
ocus of the source slit, the skewing of the boundary pattern is rela¬ 
te ely insensitive to the camera focus. The object plane of tbo 
camera can be found most readily with the aid of a'^glass scale* 
placed ,n the thermostat normal to the optic axis. A low Tower 
microscope is then focused on the imaae of a r ^ 

nstutgjhe ophcs.~\n impression of the quality of the optical 
Lyman NiXfiTll Vreiland AleVNSy\o,^N"^ •'’ailaWe from 



system may be obtained by the following means. With the thermo¬ 
stat windows removed the vertical dimension of the source slit is 
set at 0.05 mm, say, with the aid of a feeler gauge and the edge of 
the schlieren diaphragm placed just below the image of this .slit. 
On the ground glass the image of the schlieren lens should now be 
uniformly illuminated at the same intensity as with the diaphragm 
removed. At unit magnihcation a perfect image of the source slit 
would also be 0.05 mm wide, and raising the diaphragm by this 
amount would cause the image on the screen to go uniformly dark. 
Mdth good equipment the image goes dark with a diaphragm di.s- 
placement of 0.06-0.08 mm, but seldom completely uniformly. 
Repetition of this test with the windows in place, then with water 
in the bath at 0.5 C and, finally, with the cell, also filled with 
water, in position in the bath affords information as to the contri¬ 
bution of each of these elements to the total distortion. Suggestions 
as to the significance of the shadows that appear on the screen as 
the slit image is intercepted are given in reference (27) together 
with additional procedures for aligning and focusing the equip¬ 

III. Interference Methods 

Rayleigh fringes .—Measurements of refractive inde.x with the 
aid of methods based on the interference of light have long en¬ 
joyed a well deserved reputation for sensitivity and precision. 
Only recently, however, have such methods been introduced into 
moving boundary studies. The one that can be adapted to muc i 
of the existing schlieren equipment with only minor changes is a 
modification of the Rayleigh interferometer suggested by Philpot 
and Cook (42) and by Svensson (50). One form that this method 
takes is shown in Figure 9. Here A is a point source of mono- 
(;hromatic light, L is a lens that forms an image ol this source in the 
nlaiie at P, M is a plate with vertical slits that mask a boundar\ 

channel and a neighboring reference channel 
C is a cylinder lens, with its axis horizontal, that focuses the cell 
(not shown) vertically but does not affect the horizontal spacing 
of the Rayleigh fringes F that are formed at P by the vertical slits. 
If the fluid in the boundary channel is homogeneous these lunges 
are vertical, as shown in Figure 9. With 1 or more boundaries in 

this channel, however, they are warped, as f ^ 

micronraoh of Figure 10, into overlapping segments oi the com 
n cte XctRe Index-height curve that intersect diagonally 
t^he central diffraction envelope (30). Owing to the relative > giea 



Fig. 9.—-Modified Rayleigh interference method for obtaining the refractive 
, index as a function of the height in the channel. 

Iio. 10. Photoinicrograpli of Rayleigh fringes obtained on electrophoresis 
o, a human serum (1 vol serum + 5 vol buffer) Here the fringe pStemSe 
using boundjines has been enlargeil in 4 sections, the crowding of the fringes 
the albumin p^k being evident in the left-hand part. Arrows indi'- 

KaToDOMlTKIO^'b ?’■<»<«'•<» inX XHTiebrio" Oo"m 

this diplacement correaponde to a path d fereX 

Svensson camera may be used at C Tt lo i ^ ^ hilpot- 

the conventional mask of the Tis^iius ceH ^ wP 

in front of each side of the channel an 1 fl ^ contains a slit 

time, by one having a double slit for each'^swT^One^of tb*^^ 
poses the chattnel attei the other the referenc^plj.r through tt 



adjacent bath water. Each slit is 1.5 mm wide and the distance, 
6, between their centers is 6 mm. To obtain interference, the refer¬ 
ence path must contain a thickness of glass equal to that of the 
channel windows. This is done by fusing duplicates of the windows 
to the outside wall of each channel, a modification of the standard 
cell that is now available.^ The “point” source may be obtained by 
superimposing a vertical slit upon the horizontal one used in the 
schlieren method. To obtain sharp fringes, the lateral dimension 
of this source should not exceed V 4 of the fringe separation, d of 
figure 10. With an optical distance, b', of 186 cm from the double 
slit to the plate, d = 6 'X /6 = 169 m at the wavelength X = 5461 A 
of the mercury green line. The vertical dimension of the source is 
not critical. In fact, if the cylinder lens is of sufficiently good 
(juality a satisfactory source is obtained simply by rotating the 
schlieren source slit to the vertical position. 

If the schlieren equipment is provided with a Philpot-Svensson 
camera the Rayleigh fringes can also be formed in its focal plane 
with even fewer changes than those mentioned above (50). Aside 
from the modified Tiselius cell with the double slit mask for each 
channel, the only change required is the rotation of the source slit 
to the vertical position. With this arrangement the adjustment of 
the axis of the cylinder lens to the vertical is quite critical and is 
facilitated by first obtaining the fringes with a point source and 
then refining the adjustment as the vertical dimension of the source 

slit is increased. ^ 

As in the case of the schlieren method, a “base line may be 
recorded by photographing the fringes with a homogeneous solu¬ 
tion in the channel. The alignment of a fringe pattern 111 the com¬ 
parator (described on p. 83), is facilitated if the cell mask in¬ 
cludes another pair of vertical slits m the path of which a propeily 
oriented glass plate is placed. A set of vertical fringes is then 
formed at P, Figure 9, that is parallel to, but displaced ^ 1 . the 
envelope of those from the boundary channel. This set thus 
affords an invariant frame of reference for each exposure. 

Janiin fringes.-ln the 2 micro-electrophoresis apparatus no^^ 
available commercially (see Table 3) the pnncple 
interferometer has been utilized. In the Kern instrnmen (21, 31) 
monochromatic light is u.sed and the image of the channe i. 
crossed by a series of horizontal fringes at positions conjugate to 
the successive levels at which the optical path difters by 1 ware 
eng h A typi a photograph of a human .serum is reproduced in 

Fieu e 11 S^ere tL curve is that obtained hy plotting the fringe 
nurllber as ordinate against its position as abscissas. 'Ihe points of 

Manufacturing Co., 207 East 84th St.. New York 28. N. Y. 



inflection in this refractive index-height curve correspond to 
minima in the gradient curve and the data may be treated as 
described for Rayleigh fringes on page 84. The warping of the 
fringes near the walls that is apparent in Figure 11 may be the 
result of electro-osmosis, or of both this and thermal convection 
since the channel of this instrument is not thermostated. It should 
be noted, however, that this photograph is a 2.5 X enlargement of 
the actual channel, the clarity of the fringes near the wall indicat¬ 
ing the wealth of detail available with this optical system. 

In the Antweiler instrument (5) the boundary and reference 
channels are “scanned” with coherent beams of white light of 

small cross-section 0.5 X 0.05 mm. With the beams entering the 

cXcidencVaXliie " "'"'f ’ brought into 

tuiuciaence with the cross-hairs m an pvpnipp^ Kir +;!+• 

second Jamin plate with the aid of a graduated drum and screw*; 
white Wnge^agahfcentered 

Figure 11. ■ef.actue nnle.x-height curve like that of 



Evaluation .—In their ])resent state of development interference 
methods are not as satisfactory as the schlieren procedures for the 
determination of boundary position, i.e., mobility measurements. 
For analyses, however, they appear to be more precise. This 
doubtless arises from the greater accuracy with which the fringes 
can be located as comjDared with the edges of the relatively coarse 
diffraction patterns that are encountered in using the schlieren 
method. The fringe displacement at a sharp boundary such as 
that of Figure 10, h, can be measured to within 1 or 2 n. With 
d — 169 M this corresponds to a path difference of 0.01 wave. 
With X = 5461 and a = 2.5 cm, 0.01 wave is ecpial to 2 X 10“^ 
in the refractive index. 

IV. Analysis of the Pattern 

Pallern areas .—For the determination of the areas enclosed by 
the peaks in a pattern such as that of Figure 4, a, the photographic 
jdate is placed in an enlarger and the projected image of the pattern 
traced on a sheet of paper (Fig. 4, d). This sheet is then shifted 
until the lower edge of the base line exposure (Fig. 4, 6), is brought 
into coincidence with the horizontal edges of the pattern con¬ 
jugate to those portions of the channel in which there are no 
gradients. Tracing of the lower edge of b then gives the base line. 
If gradients are present throughout the channel it is necessary to 
interpolate on the pattern tracing for the same diaphragm setting, 
Ni, as that at which the base line recording began and then shift 
the paper to bring the image of the upper edge of b to this level 
before tracing its lower edge. Similarly, by shifting the sheet again 
so that the image of the graduation at Ho m c is coincident with 
this line in the tracing of a, the initial boundary position can be 

recorded as shown at in d. . 

As is evident in Figure d, there is appreciable overlapping of the 
gradients of adjacent boundaries and the division of the area is 
Somewhat arliitrary. Following the procedure 
Ivabat (54), this is done hy drawing an ordinate from the lowest 
point between adjacent peaks. With the aid of a polar 
e Keufel and Esser no. 4242, the area under each peak is de- 
te?-mined in arbitrary units. In this determination it is es^ntial 
that the graduated wheel of the planimeter move on a unifoimly 
mooth surface. Since the total area of a pattern is proportional 
to the difference in refractive index of the protein and buffer solu 
onrthit ^t at the original boundary, it should be the same for 
^atterirfrom the 2 sLs of the channel in a gi-n exp-irne^ 
If the plasma proteins are assumed to have the same p 



refraction and the apparent relative composition is all that is re¬ 
quired, the ratio of the area due to the albumin boundary, say, 
to the total area of the pattern exclusive of that due to the con¬ 
centration boundary, i.e., the 5 or e effect, is the fraction of the 
total protein present as albumin. See, for example, the last line of 
Table 4. 

If the difference, A/r,, of refractive index across a boundary due 
to the fth component is required, the relation is 

An, = Ai/abGEeEtP (2) 

in which the terms not previousl}^ defined are: At = area of f-peak 
in planimeter units; G = ratio of plate to diaphragm-travel in the 
scanning process or tangent of the angle of the inclined slit with 
the vertical in the Philpot-Svensson camera; Et = tracing en¬ 
largement, and P = planimeter units/cm^. If only 1 protein is 
present and its refracti\^e increment k is known (40), the total area 
of the pattern affords, with the aid of equation (2), an accurate 
measure of its concentration p, since p = An/it. 

TABLE 1. —Determination of the Optical Distance, h 


Cell window 

Cell window to inside thermostat window 

Inside thermostat window 

Air space in double window 

Outside thermostat window 

Outside thermostat window to diaphragm 


















h = 182.06 cm 

Channel depth.—'the channel depth a is deteimined with the aid 
of a small brass bar whose length / is slightly less than a and whose 

on1he St!*' ‘‘f' in a horizontal position 

on the stage of a microscope the bar is inserted and the difference 

a - I determined with the aid of the graduated fine adjustment by 

top winit and the inside wall 

chaimel " ^''^P^ated as the bar is shifted along the 

The optical distance “6.”— If there is nn , 

i^iiSiati t'r«-S'e f r"r 

‘ri:: r 

The factor h", the 1 nf ^b ‘ T'"’ ^d column. 

.1. ii..» Si Si,;,' 



the focal length of the 2d schlieren lens and can frequently be ob¬ 
tained from the manufacturer. Since the principal planes of most 
achromats lie within the glass, the distance from the crown of the 
lens to the diaphragm is somewhat less than h. A small glass prism 
of appropriate and accurately known power in water at 0° C is, 
however, the most convenient accessory (27) for the determination 
of b. If such a prism is placed in the thermostat with the plane 
bisecting the prism angle coincident with the object plane ol the 
schlieren camera, the resulting displacement of the slit image is a 
measure of b. This displacement may be obtained by noting the 
diaphragm positions at which the image of the prism and that of 
the remainder of the field go dark. It is also clear that h may be 
determined with the aid of equation (2) if a boundary is present 
in the channel across which the difference of refractive index is 
known. One formed between water and a 1 per cent solution of 
sucrose may be used if it is shifted into view sufficiently slowly- 
At 1° the refractive index increment for this solution is 1468 X 
10at a wavelength of 5461 A, i.e., the mercury green line (13). 

Enlargement factor s.^Both. enlargement factors of equation (2), 
i.e., Ec and Et, are conveniently determined with the aid of the glass 
scale. This is placed in the object plane of the schlieren camera 
and photographed, and the ratio of the scale interval in the nega¬ 
tive to that in the real scale, as determined with the aid of a com¬ 
parator, is Ec. To determine Et the enlarged image of a contact 
print of the scale is projected onto the scale itself. 

Computation of mobility.-The enlarged tracing of the pattern 
can also be used to determine the boundary displacements iiom 
which to compute mobilities. The initial boundary position is 
taken as the center of the schlieren band after interruption of t le 
compensation but before passage of the current As is fiom . 
comparison of Figure 4, a and c, it is incorrect to ^ 

concentration boundary does not move on passage of the cuuent 

and to use its position as that of the initial boundary. 

Tf the gradLt curve for the boundary after electrophoresis is 
symraetriLl the midpoints of a series of chords will 
line parallel to that due to ‘he gradua^on o^^ . ^ 

th! boundary is skew and its --cb 

position as “ th error with the aid 

gradient curve. This - pnndtictance of the protein 

of the plaiiimeter. If is the specific conductance P 



solution and ^ amp is ])assed for t sec in causing the boundary to 
move through a distance of A// cm on the tracing, the correspond¬ 
ing displacement, A/;, in the channel is AII/EcEt and the mobility is 





where s is the average cross-sectional area of the channel. 

Channel cross-section .—For the determination of s the asseml)led 
Tiselius cell is filled with mercury to a level somewhat above the 
junction of the center and top sections. Shifting the center section 
to 1 side then isolates its contents and permits pouring off of the 
excess mercury in the top section. The center section is next re¬ 
turned to its original position, but isolation of the mercury therein 
fiom that in the bottom section is retained by shifting this section 
simultaneously with the center one. Now, by pouring the contents 
of the 2 sides of the center section into 2 previously weighed 
beakers and reweighing, the total volume of eacdi si(ie may be 
computed with the aid of the density of the mercury at the tem- 
IDerature of the operation. An average value of s that is adequate 
loi most work is the quotient of this volume and the height of the 
section as determined with the aid of a vernier caliper This is an 
averap howm^er over the entire height, including the horizontal 
end-plates. The channels of most center sections have a reasonably 
umform (u^oss-section pppt where they pass through the slots in 
the end-platp. In careful work correction for this can be made by 
p itiactmg the vpime of a channel in these plates from the total 
(lume and dividing the difference by the total height less the 
nckness of the plates (3). The thickness, width and dept^of a 
. 0 m an end-plate can be determined by means of a micrometer 
kelei gauge and inside vernier caliper, respectively. 

nnqe analysis.-~Fi)Y the analysis of interference frino-e nattern*? 

lie -t w! 

r fhp r ! T T straight to within 0 1 

»l the .,o., d and tl.e allowable er,-or in n is 2 X 

The pattern of the bounXryTystLinecessary. 

"•ith the Ycrtieal fringes conj.fgtte to the hi 
l>uller at the top of the 

instrument. With the 1 . .• P^^‘^nel to the axis of that 

say, the reaclinr// at ^ 

IS talnilated. Since eacli fiinge 

mnX.‘ighthhat\',dei i'“‘f'’^ r"'® ''"‘“'‘''S'^ the fringe' 



tioiis between the fringes are plotted as ordinates against their 
mean position H as abscissas the curve of refractive index gradient 
versus height is obtained. This is shown in Figure 12, where the 
circles indicate the values of 1/A// for the fringes of Figure 10 and 
the full curve is a tracing of the schlieren scanning pattern ob¬ 
tained in the same experiment. From this figure it is clear that the 
2 independent optical methods give the same result. 

For an analysis it is unnecessary, however, to prepare a plot. 
The position of a minimum in the gradient curve corresponds to 
a maximum in the fringe separation, and the number j of the 

Fig 12.—Pattern of a human serum (1.2 per cent total protein concn.) ob¬ 
tained by 2 independent optical methods. 

fringe at which it occurs can be determined by inspection of the 
differences AH, or to a fractional part of a fringe by numerical 

interpolation (30). Thus, if A//i = Hj — //^-i, AH 2 — 

//,, and AHs = Hj +2 - H,^, and AH 2 is a maximum i.e AH, < 
ah 2 > Alh, the value of j at the apex of the curve of AH versus j 


(2i + D^Hi - - (i + i)ARi 

“ 2MI2 — aRi — Ai/s 

Ill Fiaure 10, tor example, the values of A//i,i,3 at j - 44 are 
466 5.32 and 403 a, re.spectively, and j„ = 
maxima occur at j = 0, 44.6, .50.4, .56.7, 

this last interval repre.seuliug the 5, re., pruleu, co ,cc d .d, u 
boundarv The i.ositions of the.5e maxima are ludK.ded l>> I ■ 
arrows in Fig 10, a. 11 all of the serum proteins have esseutialW 
“me specific ;-efractiou the apparent composition of this serum 


is 44.6/92.0 or 48.5 per cent albumin, 6.3 per cent ai-, 6.8 per 
cent a 2 -, 13.4 per cent /3- and 25.0 per cent 7 -globulin. 

If the base line fringes are not straight a refractive inde.x- 
height curve for the imperfect optical system must be constructed 
and subtracted from that with the boundaries present in the 

V. Interpretation of the Pattern 

Ideal electrophoresis .—Within the pH-stability range of most pro¬ 
teins the mass/unit net charge, i.e., the equivalent weight, is 
large in comparison with that of the buffer ions. Thus a 1-2 per 
cent solution of protein usually corresponds to a very low equiva¬ 
lent concentration of this material although it makes a major 
contribution to the density and refractive index. The conductance 
and pH, on the other hand, are determined largely by the buffer 
ions and, if these are present in sufficient concentration, provide, 
on passage of the current, a uniform electric field in which the pro¬ 
tein ions drift. In the ideal case of low protein and high salt con¬ 
centration there are then no superimposed gradients of other 
species at a boundary across which a protein disappears, and the 
patterns from the 2 sides of the channel are mirror images of each 
other, i.e., they are “enantiographic.” 

The evaluation of the apparent composition and mobilities de¬ 
scribed in the previous section assumes that this ideal situation is 
realized m practice. Actually it is not, the departure from the ideal 
in the case of real patterns being evident from the presence of the 
concentration boundaries and from the fact that the volume 
through which a boundary rises on passage of the current is gen¬ 
erally greater than the displacement of the corresponding de¬ 
scending boundary. Moreover, the area due to a rising boundary is 
usually less than that of the corresponding descending boundary 
compensation for the differences being made bv the 5 boundarv 
whose area is greater than that of the e effect. In the absence of 
immobile components the c boundary is merely a change in the 
concentration of the buffer salts, and all ion species, both buffer 

ConC Pi'oportion across the 5 boundary 

S.m,w f eiiantiogmphy would not be a, 

nice of on 01 if it were not for the fact that the disappearance of 

iuldmerb hrir ", ''“""JHi-y is accoinpa.nie,l‘l,y ad^ 

c^oeTbl Iho'vlul": I'llot 

of the apparent composition. Thus the 



question arises as to the extent the apparent composition diflers 
from the true value. 

Apparent and true composition .—Since no sufficiently precise 
measurements on protein mixtures of known composition have 
been made, a really satisfactory answer to this question is not 
available. The work of Perlmann and Kaufman (39) affoi-ds, 
however, an idea of the magnitude of the error. With the aid of a 
O.IM sodium diethylbarbiturate buffer at pH 8.6 they found that 
the apparent albumin content of a human plasma, as determined 
from the patterns of the rising boundaries, varied linearly from 
58.2 to 54.6 per cent as the total protein concentration was re¬ 
duced from 2.66 to 1.00 per cent. In the case of the patterns of 
the descending boundaries the variation was less, being almost 
within the limits of error of the measurements, but the results from 
both sets of patterns extrapolated to the presumably correct value 
of 53 per cent albumin at zero concentration of protein. 

The error also depends upon the nature of the buffer solvent, di¬ 
ethylbarbiturate being one of the best in this regard (28). Dole’s 
theory (9) has been useful in indicating the magnitude of the error 
to be expected in a given medium and is in qualitative accord with 
experiment. However, this theory does not take into account the 
effect of the small pH changes across the boundaries and in many 
cases, including serum analyses, indicates larger differences be¬ 
tween the true and the apparent composition than probably exist. 

Both theory and experiment indicate that the apparent values 
obtained from the pattern of the descending boundaries are more 
nearly correct than those from the rising boundaries. This is true 
of both the composition and the mobilities. In fact, the mobility 
computed from the displacement of the fastest descending boundary 
is correct since here a species, albumin in the case of serum a p 
8 6, disappears from a solution, the protein solution whose con¬ 
ductance is known. Across all other moving boundancs in the 
system a species disappears from a solution that has been cie. ^. 
by the electrophoresis and in which the conductance dilfeis shg i y 

from that of the original protein solution (51). 

TfMitv vs. pIL-ln view of the foregoing coiisideraUons the 

idl-mobilfty curve of a protein, and hence its isoelectric pH, shou 
l^e d";Ld. if possiWe. upon samples of 
Moreover, concentrations of not more than a 

2.«,, -j —tr"S,rr 

‘“Intddilirt:- the departures from ideal behavior .nentioned 



above there are those encountered with mixtures in Avhicli inter¬ 
action between the components occurs. Although the considera¬ 
tion of such systems is beyond the scope of this chapter, references 
(4), (25) and (45) are to work in this field. 

VI. Preparation of Solutions 

Buffers .—In electrophoresis it is frequently necessary to pre- 
I)are a variety of buffer solutions and it is well for the investigator 
to be familiar with the most commonly used systems. These are 
listed in the 1st column of Table 2 in the order of their pK' values 
as given in the 2d column. The pK' of a buffer system is the pH 
of a half-neutralized solution of the weak buffer acid, or base, at a 
gi\^en ionic strength. In Table 2 the values of column 2 are for an 
ionic strength, fx, of 0.1 and a temperature of 25 C. The classifica- 

TABLE 2.—Buffers for Use in Electrophoresis 


Buffer System 



M = 0. 1 

25 C 

Glycine, HCl 


/3-alanine, HCl 

Acetic acid, NaOH 


NaH2P04, Na2HP04 


Diethylbarbituric acid, 


Glycine, NaOH 


/3-alanine, NaUPI 

* Extrapolated, t pK' (0®) — i)K' (2.5°) 



M = 0 

25 C 



n = 0 

0 C 


































tion of tlie buffers, as listed in column 5, is on the basis of the 
charge type of the acid form as suggested by Br0nsted (7). If the 
buffer IS of the uncharged- or anionic-acid type the pH of a given 
solution increases slightly with dilution, as is indicated by the pK 
values at zero concentration, column 3. 

Since it IS inconvenient to thermostat the glass electrode and 
vacuum tube potentiometer used for most pH measurements, 
hese aie generally obtained at a room temperature of 20 to 25 C. 
n the case of a buffer whose pH is less than 7 the value thus ob- 
tamed is not very different from that at 0 C, the temperature of the 
electrophoresis. Alkaline buffers, on the other hand have ao- 
preciahle temperature coefficients. This is shown in Table 2 where 
the change m pK from 25 to 0 C, columns 3 and 4 is a close an 
proximation to the shift in pH over this temperature ’interval Thus 
a glvcme buffer of pH 9.60 at 25 C will have a pH ono",5 at 0 C 



Owiiijy to the hygroscopic nature of many buffer salts it is gen¬ 
erally best to prepare the solution from standard solutions of s 
strong acid, or alkali, and tlie weak buffer electrolyte, lu the case 
of uncharged acid systems, B moles of sodium hydroxide ar(‘ added 
to A moles of the weak acid, where A > B. As a close approxima¬ 
tion the pH is given by the relation 

pH = pK' + log + '‘’‘5 SS) 

and the ionic strength is the concentration of the sodium ion if this 
is large in comparison with that of the hydrogen ion. Thus, a 
solution containing 0.1 mole of sodium acetate and 0.01 mole ot 
acetic acid per liter is O.IM and O.OIM in these 2 constituents, 
respectively, and has a 

pH = 4.64 + log = 5.64 

and an ionic strength of 0.1. As another example, suppose the 
composition of a 0.1 y. glycine buffer of pll 10.00 is required. 
Equation (4) is rearranged to 



^ = antilog (pH - pK') = antilog 0.33 = 2.14 

Since (salt) =0.1 the acid concentration must be 0.0467 and would 
require, per liter of solution, 0.1 mole of sodium hydroxide and 
0.1467 mole of glycine. 

In the case of phosphate buffers equation (4) may be used it 
(acid) = MNaH,po 4 and (salt) = MNa 2 HP 04 but the ionic strength is 
U = MNaH,po. + 3MNa.HP0.. All of the othei' buffers of Table 2 are 
monovalent and afford a maximum buffer capacity for a given 

ionic strength. , . , t m„Kio 9 

Selecting a &M^er.-Conspicuous by their absence from lable 2 

are monovalent buffers for the physiologically important lange 

from pH 5 to 8. Several systems, e.g., cacodyhc acid and the su - 

stituted phenols, are available but must be h cau .on 

because of their tendency to denatuie proteins. Othe s juc 

the substituted glyoxalines (20), are not 

others eg. formic, lactic, citric, phthalic and bone acids, are 

omitted because of their tendency to form 

teins. although it cannot be presumed that those of Table 2 

'lirllC^ictSTiuffer system is used to cover a pH range 
of a'boTtI r! on either side of its pK' value- In e^ctr^ 
however, the effect of high concentrations 
trolyte upon the temperature of maximum density ol the 



should be considered, as is shown later (p. 95). Thus a 0.1 m ^ alanine 
buffer at pH 3.64 would be preferable to an acetate buffer of the 

woidd be IM. Another consideration in the choice of a buffer is 

scending boundaries are enaiitiographic. If the proteins are 
present as anions, enantiography is favored by the use of a buffer 
of the uncharged acid type; if present as cations, a cationic acid 
type is preferred (28). A high buffer capacity is not always essen¬ 
tial, however, and it is sometimes desirable, especially in obtain¬ 
ing pH-mobility curves, to use low concentrations of the buffer 
electrolytes and build up the ionic strength by addition of a neu¬ 
tral salt such as sodium chloride. 

Stock solutions. —Coin'enient stock solutions for use in the 
preparation of buffers are 2M sodium hydroxide, 2M acetic acid 
and 2AI hydro(*hloric acid. Since it determines the ionic strength in 
many instances, the 2M NaOH is the most important of these 
reagents and is prepared as follows. One lb of reagent grade NaOH 
(from sodium) is dissolved in an ecpial weight of water. On stand¬ 
ing the sodium carbonate present as impurity, being insoluble in 
this 50 per cent solution, settles and the clear supernate is decanted 
into a stoppered, paraffin-lined flask. The density of this solution is 
1.525, and 117 ml is required for 1 1 of 2 -b M solution. The exact 
concentration is determined by titration against potassium acid 
phthalate with phenolphthalein as indicator and is then adiusted 



Dialysis .—As noted on page 63, the protein solution is prepared 
b}’- dialysis against the buffer solution. This is done with the aid of a 
bag made of Visking cellulose tubing, a long one of small bore being 
preferred. For example, after an 8 in. section of in. tubing is 
made pliable by soaking in water, one end may be twisted and a 
knot tied therein. After testing for leaks by filling the bag with 
water under gentle pressure it is emptied and 12-15 ml of the pro¬ 
tein solution introduced, together with a glass bead. The bag is 
then closed by tieing a knot in the upper end, after which it is 
suspended in about 250 ml buffer in a stoppered, footed cylinder 
and placed in a refrigerator. Inversion of the cylinder a few times 
during the 12-24 hr period required for attainment of the Don nan 
equilibrium piovides some agitation. The buffer in the cylinder is 
replaced by a fresh portion for a 2d period, after which the bag is 
transferred to the remainder of the buffer solution, 1.5 1 if a 2 1 
sample was prepared originally, for a 3d period of 12-24 hr. This 
buffer dialyzate is then used in forming the boundaries and for 

filling the electrode vessels. 

Although the foregoing dialysis procedure eliminates virtually 
all diffusible solutes except the buffer electrolytes, this is not always 
essential. Dialysis overnight against the 1.5 1 of buffer required 
for the cell and electrode vessels is frequently adequate since the 
foreign ions are then present on both sides of the initial boundaries 
at low concentrations and do not give rise to moving salt boundaries 
of appreciable magnitude. The time can be reduced by dialysis 
(43) at room temperature. Continuous agitation also hastens the 
process somewhat, although the rate appears to depend mainly on 
diffusion through the cellophane. Most tubing has a wall thickness 
of 0.02 mm when dry, and material having walls thicker than this 

should be avoided. , , • ♦ . 

Serum or phrsma is usually diluted to a total prote.u coueenlu - 

tion of 1-2 per cent foi- electrophoresis. llu.s may be done eil in 

before or after dialysi.s, the diluent in the first ca^ '"'.rsenimf 
vvherea.s in the second case it is the buffer dial.ysate. Seiums o 
plasmas that have been diluted with the buffer mt not dialyzed 
S it should not be used. As a result of the <hi^;-u o - 
the concentration of protein in the bay ehaiiKes slishth duiiii„ 
dialysis, a fact that is sometimes overlooked. 

VII. Measurement of the Conductivity 

Comluctuitii fc/l.-rhe conductance of the protein .solidion ma.v 

he wiiit ui'i o{ 

cell developed by Shedlovsky (44). Hu. lUdioi 



(Fig. 13) are hollow truncated platinum cones e-e, the outer sur¬ 
faces of which are sealed to the glass and make contact with mei- 
cury in the tubes t-l by means of platinum wires. The imier sur¬ 
faces in contact with the solution ai'c platinized (44). Owing to the 
relatively unimpeded flow when the cell is filled by means of the 
tubes/-/, there is little tendency for air bubbles to be trapped, an 
advantage in the case of protein solutions. The glass rod r-r 
atfords mechanical strength and is used to suppiort the cell in the 
1.9 1 Pyre.x Thermos vessel #8621 that serves, when filled with 
shaved ice and distilled water, as a thermostat lor the conductance 

A cell of this type requires 3 ml of solution and has a constant, 
/y, of about 16. This constant is detei-mined with the aid of a solu¬ 
tion of known conductance; e.g., one containing 7.4789 g of potas- 

ofTen!;!';');''veiKht, i,. ai,. 

sistonee ofohms, it« specific 

the l)i i<lse described below is most sensitive F ^n' ‘ 

cell having a constant of 1-2 nw / ^ solutions a 

13 may be shunted will, a 

termined value /?, of sav 10 099 \ rf accuratelv de- 

-^.stance, that .ut 

'' ... c ,!:!:7^;7'sirt!:e“ ;ho 



arms are fixed at 1,000 ohms eacli, the cell resistance, when 
balanc(;(l, is fj;iven directly to the nearest ohm by the setting of the 
decade dials and to the nearest 0.01 ohm by the graduated 1.1 
ohm rheostat ll in the ratio box. The wiring diagram of the asso¬ 
ciated oscillator and amplifier is shown in Figure 14. A frequency of 
about 2,000 cps is used and, by shielding the components inde¬ 
pendently as indicated by the dotted lines in the figure, it is pos¬ 
sible to house the entire bridge in a 19 X 18 X 13 V2 in. Par-metal 
cabinet without the amplifier picking up a signal directly from the 

The balancing procedure is as follows. With the switch thrown 
to the resistance arm and the capacities Ci and ci are adjusted 
to give silence in the phones. The switch So is then thrown to ground 

to all 


L....... . V. 

Fig. 14 .—Coaductivitv bridge. .411 fi.xed resistors are 1 W; values in megohms. 
All fi-x’ed condensers are 400 v; values in microfarads, a = 6 y storage battery 
6 = 90 V “B” battery. c„ c, ca = 30 350, 350 respectively variable an 
condensers, rf = 30 nblUhenry radio frequency choke Bud^^^ vM inir/vri.l 

60 millihenry radio frequency choke, Budd #CH1219. t - 1 /a v Mallory gii 1 
IWe P = crystal phones. Brush model A. s. = dpst and spdt switche^ 
= input transformer, General Radio 578A (detector terminals of the ratio 
box may also be connected directly to the amplifier, omitting this transformer). 

and silence is again obtained by balancing the earthing capacity 
CAP and the earthing resistance G of the ratio box. 1 he capacity Cs 
shunts that of the box and can also be used in making this balance. 
Now with S2 returned to i) a new bridge balance is found it 
neceiary, this process is repeated until silence is obtained with S2 

S'cfrclucuni is measured at 0° C, the mobility that is 
computed with the aid of equation (3) is the value aUhis 
tiiie'^although the boundary displacement occurs at a someii . 
Sertemperature. This is due to the fact that the coiiductaiice 
!s sens" the temperature whereas the displacement changes 

""It is’^dsif good practice to determine the conductance of the 
biiffei dialysate since a comparison with that ol the protein solu- 



tioii iiidicatps the effectiveness of the dialysis, the values for the 2 
solutions I’eaching a small but constant difference as the Donnan 
equilibrium is attained. Owing to the precision with which con¬ 
ductances can be determined, such a measurement affords a check 
on the care exercised in the preparation of the buffer solution. 

VIII. Control of Convection 

Convo.clive circulation .—By convection is meant the circulation 
of solution in the channel that results from (a) the horizontal 
density gradients accompanying the dissipation of the heat that is 
generated by the passage of the current, i.e., the Joule heat, (6) 
electro-osmotic streaming, (c) the development of layers of solu¬ 
tion less dense than those above or more dense than those under¬ 
neath and (d) possible drainage effects due to the layer of im¬ 
mobilized solution at the wall of the channel. This last effect ap¬ 
pears to be a source of convective disturbance only in the case of 
abnormally viscous solutions, such as those of the nucleic and 
hyaluronic acids, and is apparent in the distortion of the boundary 
that occurs on shifting it in the channel. 

In electrophoresis in the Tiselius cell it is safe to assume that 
some circulation always occurs in the conducting, homogeneous 
columns of solution between the boundaries. The problem, then, 
is to choose conditions which reduce this circulation to such an 
extent that the vertical density gradients at the boundaries can 
stabilize them against the disturbing effects of the convection. 
The requirement for gravitational stability, namely, that the 
ight solution must always be above the heavy one, represents a 
limitation to the Tiselius method that is not present in packed 
channels where the packing material provides interstices in which 
separations can occur without regard to this requirement. Al¬ 
though electrophoresis in packed channels, e.g., paper strin 
lonography (11) lonophoresis in silica- (8) and agar-gels (38) and 
thei packings (49), is now being investigated for qualitative 
analyses and preparative purposes, it will not be treated here. 

e ^-gobuhn disturbance that is encountered in work with 

stability At”(IT'l " T\ «'-«vitational in- 

vect offrenLntW T * ''“'maary localized con- 

pSern. ■ ' sharp “spike” in the 

boundary work is not yet clear. In 1940 tt ktrMrKleltT''l7 
urn,shed the autho,- 2 center sections i.leni, 10 ^,^th iV of P ''' 

>. a, except that the channel widths we ' 



spectively. In each of the 3 channels thus made available, a 1.16 
per cent solution of ovalbumin in a 0.1 /x sodium phosphate buffer 
at pH 6.8 was allowed to migrate about 6.7 cm at an electric field 
strength E of 7.5, and also at 15 v/cm. The essential portion of 
each pattern of the rising boundaries is shown in Figure 15. Since 
the best pattern was obtained under the conditions usually em¬ 
ployed, namely, in the Tiselius cell at the lower field strength, 
further work with the narrow channels was discontinued. In 
view, however, of the recent introduction of microcells with chan¬ 
nel widths as low as 1 mm, the patterns of Figure 15 acquire a new 

The most striking feature of the patterns is the progressive de¬ 
terioration, at both field strengths, of the resolution of the Ai, 

7.5 volts/cm. volts/cm. 

15^—Patterns of the rising boundaries of ovalbumin in channels of differ¬ 
ent width, w, and at different field strengths, E. 

and As components of ovalbumin as the width of the channel is de¬ 
creased Moreover, in each channel the maximum refractive index 
gradient is lower at E = 15 than at E = 7.5. Although electro- 
osmotic streaming may be a factor in the narrow channels, it does 
not appear to be directly responsible for the interior resolution 
since the “stationary” concentration boundary in each experiment, 
except in the Tiselius cell at ^ = 15, was unaffected. When a 
boundary is shifted at a given rate by means of the compensator it 
spreads more rapidly in a narrow channel than m ^ ^ 

possibly to relatively slow “drainage” of the immobilized soluti 
in contact with the wall. Although a boundary moves m an electric 
field as a result of ion migration, and does not involve dispHce- 
ment of the solution in bulk, as during compensation the effects o 
the resolution appear, superficially at least, td be similai. ,. 

EffTofJouUhea^^^^^^^ the other hand, the irregular grafts 
in the 3 mm channel at 15 v/cm (Fig. 15) are typical of those en 



countered when excessive field strengths are used in the Tiselius 
cell and doubtless result from inadequate dissipation of the 
Joule heat. No such effects were observed in the narrow channels 
at either field strength, indicating their superiority as regards 
heat dissipation. In channels much wider than 3 mm, e.g., those 
used in the so-called separation cells, the dissipation of the Joule 
heat becomes increasingly difficult. Although electrophoresis is 
useful for preparative work in milliliter quantities and is in¬ 
valuable in piloting the various fractionation procedures that are 
adaptable to large volumes, it does not lend itself readily to direct 
separations on a gross scale. 

The convection resulting from the Joule heat has always been a 
serious problem in the electrophoresis of proteins, and one of 
Tiselius’ contributions was the minimizing of this by use of a 
thermostat regulating near the temperature of maximum density, 
TMD, 4 C in the case of water. In the neighborhood of the TMD 
the density is practically independent of the temperature, and al¬ 
though the temperature gradients in the conducting solution are 
much the same as at other thermostat settings they are not ac¬ 
companied by the differences of density that produce the con¬ 
vective circulation. 

The optimum thermostat setting is not 4 C, however, for 2 
reasons. First, allowance must be made for the fact that the mean 
temperature of the conducting solution is above that of the ther¬ 
mostat. Second, and most important, practically all solutes lower 
the TMD of water. Thus the TJMD for O.IM sodium acetate is 
2.85 C and would be still lower for an acetate buffer of this ionic 
strength. For the solutions most commonly used in electrophoresis 
the optimum thermostat temperature is near 0 C. A setting of 
0.5 C is convenient in that at this temperature there is little 
tendency for ice to form on the cooling coil and bridge over to the 
th^-mostat, thereby transmitting vibrations from the compressor. 

The correlation between disturbances due to the Joule heat and 
the slope of the temperature-density curve for the solution is close. 
Moore (36) found, for example, that the disturbances at the serum 
protein boundaries in a urea-rich buffer at 0 C were not unlike 
those occurring in the urea-free buffer at 17 C where the slope of 
the temperature-density curve for water was similar to that of the 

STror"" f The convection caused the 

pattern of each boundary to split into several sharp spikes whose 

C ^d^witlf changed erratically with time. Similarly, at 0.5 
C and \Mth the same current m both experiments, the author has 


0 C, than m a O.IM solution of the same salt, TMD = 



+2.7 C, where 8 times as much heat was developed due to the 
higher resistance. 

Temperature control .—In most installations the thermostat 
water is cooled with the aid of a refrigeration unit whose cycling 
is controlled by a mercury thermoregulator of the sealed type to 
avoid condensation of moisture at the contact. Although adequate 
for most purposes, regulation closer than a few hundredths of a 
degree is not practicable with this method since the compressed gas 
that is present when the compressor stops continues to pass the 
expansion valve and lower further the coil temperature. For 
improved regidation the compressor can run continuously and the 
excess cooling can be balanced by a thermo-regulator-controlled 
heater. A preferred alternative, however, is a refrigeration system 
provided with a solenoid vah^e between the expansion valve and 
a “surge” tank (12). The thermoregulator controls the solenoid 
valve while the pressure in the surge tank controls the cycling of 
the compressor. 

It is essential that both the refrigeration unit, including the 
cooling coil immersed in the bath liquid, and the motor-driven 
stirrer be mounted independently ol the thermostat and optical 
system to avoid shaking these components. Excessive vibration 
of the coil or stirrer can be transmitted through the bath water 
and should also be avoided. It is preferable to locate the entire 
equipment on a firm support such as a ground floor or, if an upper 
floor, near an outside wall. 

IX. Electrodes and Electrode Vessels 

To keep the products of the electrode reaction from enteiing the 
channel, relatively large volumes of buffer are interposed. Although 
the 700 ml volume of each electrode vessel of Figure 2 is adequate 
for protracted electrophoresis, in experiments lastmg more than a 
day the neoprene sleeves tend to lose their elasticity at the low 
temperature. In such cases they may be tied or, preferabljq re- 
iilaced by ball-and-socket joints (57). A more compact electrode 
vessel (26) that is coming into general use for relatively s loi 
experiments, e.g., the analysis of serum, is shown in Figure 16 and 
:i^pi^-places^’he top section of the Tiselius cell. +« ele^ 
CUDS C which are removable, serve to keep the 5 ml of IM sodiu 
cldoride solution bathing the electrodes froin falling 
iiel. Moreover, all rubber connections are elniniiated and Ifie 

lieavy duty silver-silver chloride elec- 
tro^des, /and E' of Figure 2, can be “formed” to have a capacity 



of about 1 amp-hr and can thus carry a 50 ma current for 20 hr. 
This capacity, combined Avith mechanical strength, is attained by 
winding (23) a flat and corrugated strip of 28 gauge sheet silver 
into a tight spiral (Fig. 17, a). The ends of the spirals are anchored 
with silver screws to a hollow silver core, into which the silver tube 
T is threaded, as shown in cross-section in Figure 17, b. The insu¬ 
lating glass tube S is then sealed into the core with Pyseal. In 
the case of the light-duty electrodes, E of Figure 16, the Pyseal is 
eliminated by threading both ends of the silver tube, T of Figure 
17, c, and clamping the glass jacket S between neophrene washers 

Fig 17 electrode vessels. 

g ). Construction of silver-silver chloride electrodes. 

t^^oreover, these electrodes are made from n 3/ 

Ihe reaction taking place at the cathode or negative terminal is 
AgCl -f- electron —► Ag Cl~ 


i.e., 'Paetiori i.s the reverse. 

Ag + uj- 

AgCl + electron 

0-* I Ull 

-e the silver chloride has a larger volume than the silver from 



whicli it is formed, the compact structure of the original metal 
(.loes not favor a high capacity. Before the electrodes are placed in 
service they should, therefore, be “formed” as follows. 

New electrodes may be cleaned and very lightly etched by 
momentary immersion in dilute nitric acid. After rinsing they are 
then made the anode and cathode, respectively, in an approxi¬ 
mately IM solution of hydrochloric acid and a current of about 20 
ma ])assed. Hydrogen gas is liberated at the cathode and an ad¬ 
herent brownish-purple deposit of silver chloride forms at the 
anode. After a time gas, mostly chlorine, begins to be liberated at 
the anode also. When this occurs the connections are reversed and 
silver chloride is deposited on what was originally the cathode until 
it in turn begins to gas. The connections are again reversed and 
another cycle is begun. This is repeated until the desired capacity 
is attained. It will be observed that with each c.vcle more snlvcr 
chloride is formed at the anode before gas evolution begins her(^ 
than during the previous cycle. Moreover, as the deposits oi 
“available” porous silver on the one electrode and of silver chlo¬ 
ride on the other are thus built up, heavier currents can be carried 
without loss of active material. The currents should never be of 
such magnitude, however, as to produce blisters. 

If not abused in service, properly formed electrodes will last 
almost indefinitely. Although procedures differ in the various 
laboratories, we use one electrode as cathode, say, until the silver 
chloride is nearly exhausted and then restore the deposit by elec¬ 
trolysis, as in the initial forming of the electrodes, outside of the 
mo\dng boundary cell. When not in use the electrodes are im¬ 
mersed in distilled water to facilitate the dislodgment of air 
bubbles that may be trapped on transferring to the buffer solution 

in the cell. 

X. Preparative and Countercurrent Electrophoresis 

Many naturally occurring protein mixtures possess 1 or more 
activities of biologic interest and it is frequently desira) e ^o 
identify the component with which the activity is associated. H e 
use of electrophoresis for this purpose is now a routine proceduie 
that may be described with the aid of Figure 18. Here the veitical, 
Honai and horizontal cross-hatching represents the proteins 
A B and C, respectively, whose mobilities m a given bu ei aie 
u > u > wc > 0* The initial boundaries are shown in Iiguie 
IR a On puS^ge of a enrrent in the direction indicated 3 honnd- 
.fies wi rparate and rise (Fig. 18, 6) to the leve s /u, K and 
; tlJ caSdc side while the same number w.U separate and 



(l(^scoi)(l jji tlio anode side. Pure A will then be present between the 
2 leading boundaries in the left-band channel and pure C between 
the 2 slowest boundaries on the anode side. 

Recovery of samples .—One procedure for the recovery of a 
separated fraction for as.say follows. If the left-hand channel is 
open to the atmosphere, a thin-walled glass capillary, attached to 
the empty syringe of the compensator with small bore rubber 
tubing, is lowered until its tip is just above the boundaiy at Ar. 

ith the aid of the compensator the solution of pure A is draAvn 
slowly into the syringe until the boundary at hj, falls to the level 
of the tip. Similarly, a mi.xture of A a?id H would be obtained bv 
inserting the tip to the level he and withdrawing the solution l)e- 
tween this boundary and the one at //„. Although suital)le foi- in- 

lUG. 18.— I.leuI electrophoresi.s of a protein mixture. 

Sv', SKt 

li,c b„ttc,m sect,-,,,, „f the cell t„ onAi.fe " 

As IS evident from Fi'uire 18 h nne siae. 

mobility, is still mixed'^with \ in tl "^^e™idiato 

with C h, the Otl? le E ti e , r' 

be recovered a„d used for the I t ^ 

Small quantities of this comDouen't I ' " i ^ 

somewhat more readily us follows ('il) 'v'ft ,''® ''btained 

ot pure A. the up,,erA,,a.,,da,w tut elthsul"'f 

Passed u. a direction opposite to that used i,i tie tlginarCt 



tion, B will lag behind A as the species descend and some separa¬ 
tion will occur before the boundary due to A enters the bottom 
section. Similarly, small quantities of B could be separated from 
C in the right-hand channel. 

Countercurrent electrophoresis .—It is also evident from Figure 
18, h, that if electrophoresis is continued as indicated the bound¬ 
aries will migrate into the top and bottom sections, respectively, 
before any large proportions of A and C have been separated. 
Movement of a boundary into the bottom section always results 
in convection and must be avoided. If, however, the slowest 
boundary is given an apparent velocity of zero, as is indicated in 
Figure 18, c, while the leading boundary moves through the 
height of the channel, much larger proportions of A and C could 
be separated. Actually a boundary can be given any apparent 
velocity desired by injection, or withdrawal, at the proper rate, 
of buffer into the closed side of the apparatus with the aid of the 
compensator. The movement of the boundaries against a flow of 
solution in the opposite direction, whereby they are kept in the 
channel of the cell, is called countercurrent electrophoresis. It has 
the effect of extending indefinitely the height of the channel and 
is used not only in preparative work but also in the resolution of 

protein ions of slightly differing mobilities. 

In some instances an early form of the liselius cell (Fig. 1, c), 
in which the center section is replaced by 2 identical short ones ol 
the same total height, may be used to advantage in preparative 
work. Thus if the countercurrent electrophoresis indicated in 
Figure 18, c, were carried out in this cell, the plane of separation 
between the 2 sections would be at P and shitting the top centei 
section would isolate a solution of pure A in the left-hand leg from 
which it could be readily removed on dismantling the cell. 

Umitations.—k comparison of actual patterns of P*'® 
mixtures with the diagrams of Figrne 18 w.ll the iac 

that the system represented here has been idealized^ ^•'us 
boundaries spread as they move and the Patterns t™" * m,- 
sides of the cfiannel are not mirror images of each other. In 
current electrophoresis the most serious difficulty, * 

mi e ce of the concentration boundaries, the so-called and < 
Ss that remain, on passage of the c-rent, ea^n lahy sU-_ 
tionarv at the levels of the original boundaries. M th a counte 
the'rbi'iuidarv is carried iulo the top section wlnle the move¬ 
ment of the 5 boundary into the botloiii ««d'on iiiitiatei, come ^ 



protein solution following dialysis. If any appreciable quantity of 
material is to be prepared by repeated countercurrent electro¬ 
phoresis, this procedure will save time by affording the optimum 
3 ueld from each batch. 

XI. Micro-Electrophoresis 

Commercial equipment .—-Although the procedures described 
here have been restricted to those for the Tiselius cell, it is clear 
that similar ones would be applicable to the micro- and semimicro¬ 
cells. Moreover, an apparatus accommodating the standard cell 
could be adapted for use with the smaller ones. In Table 3 some of 

TABLE 3.—Commercial Electrophoresis Equipment 


Optical System 

Aminco Standard 

Aniinco Portable 









Schlieren cylinder 
Schlieren cylinder 
Jamin interferometer 
Schlieren cylinder 
Jamin interferometer 
Schlieren band, scanning 
and cylinder 
Schlieren cylinder 
Schlieren cylinder 
Schlieren band, scanning 
and cylinder 
Schlieren cylinder 





























































+ ^'^"V™®''oceus also available with a depth of 15 mm 
thl to a 10 mm depth since tlie light is reflected back through 

1. American Instrument Co., Silver Spring, Md 

6 LKB Cm,%^9"EasV8nrs? Ne'w Yolk 28 N Y 

7 12035, Stockholm 12, Sweden ' ‘ 

.noAt, Ca^^ Associates, now associated with Specialised Instruments Corp., Bel- 

q’ Norwalk, Conn. 

9. Strubin and Co., Basle, Switzerland. 

the availalile instruments are listed toaothor with i 

systems employe<l, the dimensions of Urn ^ haLe »nd H 

rnierocells of flio \n+p- u i i- -jlmei nistniments and the 

“it'd C 



optical sensitivity, respectively, has already been discussed 
(p. 93 and equation (1)). The height, on the other hand, is an 
important factor in determining the resolving power of a given cell. 
Thus the height required for the resolution of the components of a 
4X diluted human serum may be inferred from the patterns of 
Figure 19 that were recorded, during electrophoresis in O.IM 
sodium diethylbarbiturate of pH 8.6 at 4.42 v/cm, at the intervals 
given in column 2 of Table 4. To conserve space, only the patterns 
of the rising boundaries are given and only the final one, g, con¬ 
tains the complete albumin peak. In column 3 of Table 4 are 
given the displacements of this boundary in the patterns of Figure 

19, whereas the values of column 4 are the minimum heights that 

include all of the refractive index gradients. 

Resolving power.—li by resolution is meant the appearance of a 

minimum in the gradient curve, i.e., a point and 3 

curve of refractive index versus height, then only the aa- and /3 
XbuUn boundaries are resolved in the patterns a and 6 
of Fiaure 19 In c a minimum between the ■y-globulm and ou 
fry a^Lr!; thus permitting the evaluation of ‘he —trat.on 
of this component. Separation of arglobulm fiom albumii 
fst to oZr and. although evident in e, is not well defined unt.l 

^^A^scT'incluS’ ?n Table 4 are the areas, in arbitrary units, of 



if the electrophoretic pattern must be obtained quickly, use of the 
small cells is indicated. To retain the stabilizing effect of the verti¬ 
cal density gradients at the boundaries, it is probably better to 
examine a limited quantity of material in a small cell at a moder¬ 
ate concentration than to dilute it to the volume required for the 
standard one. However, resolution in the short cells is inferior to 
that in the tall ones. Even when allowance is made for the diffusion 
that occurred on shifting the initial boundary in obtaining the pat¬ 
terns of Figure 19, and for the rather low potential gradient em- 

TABLE 4. —Dependence of Analysis of Serum Upon Resolution of 








Area in Planimeter Units 





























































Apparent composition 64.3 4.6 6.8 11.4 12.8% 


ployed, it is probable that in a channel much shorter than 50 mm 
the ai content would be uncertain and in one less than 20-30 mm 
high the division of the area due to the unresolved y~d boundarv 
would be difficult. 

XII. Conclusion 

Although more than a decade has elapsed since Tiselius adapted 
the moving boundary method to the electrophoresis of proteins 
neither the apparatus nor the technique has crystallized into a 
rigid form. This is doubtless indicative of the vitality of the 
method and of the ever-increasing field for its application Al- 
though the procedures suggested in this chapter are ones that 
experience has showtr are capable of yielding precise results, it 
«111 be surprising, indeed, if they do not undergo further refine¬ 
ment and change. The emphasis has been, however, on techZue 
rather than application. In an effort to compensate for this^in 

encL containing extensive refer¬ 

ences to the literature have been included in the bibliographv and 

son’s thest(48)'' ‘o S^ens- 



Note. —This section was reviewed by Duncan A. MacTnnes, Gertiude 
E. Perlmann, Theodore Shedlovsky and Lena A. Ijewis. 

* 2 . 




0 . 

10 . 

11 . 

12 . 




10 . 






Abramson, H. A.; Moyer, L. S., and Gorin, M. H.: Electrophoresis of 
Proteins (New York: Reinhold Publishing Corp., 1942). 

Alberty, R. A.; Introduction to electrophoresis; I Methods and calcula¬ 
tions; II Analysis and theory, J. Chem. Education 25: 426-433 and 

Alberty, R. A., and King, E. L.: Moving boundary systems formed by 
weak electrolytes: Study of cadmium iodide com])lexes, J. Am. Chem. 
Soc. 73; 517-523,1951. 

Alberty, R. A., and Marvin, II. H.: Protein-ion interaction by the mov¬ 
ing boundary method, J. Phys. & Coll. Chem. 54; 47-55, 1950. 
Antweiler, H. J.: Quantitative mikroelektrophorese, Kolloid Ztschr. 

Armstrong, S. H., Jr.; Budka, M. J. E., and Morrison, K. C.: Prepara¬ 
tion and properties of serum and plasma proteins: XI Quantitative 
interpretation of electrophoretic schlieren diagrams of normal human 
plasma proteins, J. Am. Chem. Soc. 69: 416-429, 1947. 

Brpnsted, J. N.; The acid-base function of molecules and its dependency 
on the electric charge type, J. Phys. Chem. 30; 777-789, 1926. 
Consden, R.; Gordon, A. H., and Martin, A. J. P.: lonophoresis in sdica 
jelly, Biochem. J. 40; 33-41, 1946. 

Dole, V. P.: A theory of moving boundary systems formed by strong 
electrolytes, J. Am. Chem. Soc. 67:1119-1126, 1945. i -to 

Donnan, F. G.: The theory of membrane equilibrium, Chem. Rev. 1; 73- 

90,1924. ... 1 , / n 

Durrum, E. L.; A microelectrophoretic and microionophoretic tech¬ 
nique, J. Am. Chem. Soc. 72: 2943-2948, 1950. at a • 

Costing, L. J.; Hanson, E. M.; Kegeles, G., and Morns M. S.: Equip¬ 
ment and experimental methods for interference diffusion studies. 
Rev Scient. Instruments 20: 209-215, 1949. 

Costing, L. J., and Morris, M. S.; Diffusion studies on dilute aqueous 
sucrose solutions at 1° and 25° with the Gouy interference method, 
T Am Chem Soc 71:1998-2006,1949. 

HamtX’n. S., and Owen, li. B.: The Pkfsiml Chemetry of El^lrdyltc 
Solutions (New York: Reinhold Publishing Corporation, 1943). 
HiShcocTDavid L, and Taylor, Alice C.: The standardization of hydro- 
gen ion determinations: I Hydrogen electrode measurements ^Mth a 

liquid junction, J. Am. Chem. Soc. 59: ^^^2 , ' , ^ j 

Intematimal Critical Tables (New York: McGraw-Hill Book Co., Inc., 

Jotaon/p',' ind^sfootet*!. M.: An 

aries in diffusion ex])eriments, J. Ihjs. & ooii. 

1947. , XT I .. A . r.lvoxalines- The determination 

Kirby, A. H. M., and Neuberger, A., tilyoxannes. 

. A.t.Ti»k indicales review with extensive references le the literature. 



of their pK values and tlie use of their salts as buffers, Biocheni. J. 32: 

21. Labhart, H., and Staub, 11.: Mikro-I'dektrophorese, Helvet. ehiin. acta 


22. Ijongsworth, L. G.: A niodifi<-ation of the schlieren method for use in 

' clc(drophoref i(! analysis, J. .\ni. Clheni. Soc. 61: 529—530, 1939. 

23. Longsworth, L. G., and ]\Iaclnne.s, D. A.: Electrophoresis of proteins 

by the Tiselius method, Chem. Rev. 24: 271-287, 1939. 

24. Longsworth, L. G.: Recent advances in the study of proteins by elec¬ 

trophoresis, Chem. Rev. 30: 323-340, 1942. 

25. Longsworth, L. G., and Macinnes, D. A.: An electrophoretic study of 

mixtures of o.xalbumin and yeast nucleic acid, J. Gen. Physiol. 25: 

26. Longsworth, L. G.: A differential moving boundary method for trans¬ 

ference numbers, J. Am. Chem. Soc. 65: 1755-1765, 1943. 

27. Longsworth, L. G.: Optical methods in electrophoresis, Ind. & Eng. 

Chem. (Anal. Ed.) 18: 219-229, 1946. 

28. Longsworth, L. G.: The quantitative interpretation of electrophoretic 

patterns, J. Phys. Chem. 51: 171-183, 1947. 

*29. Longsworth, L. G.: Moving boundary studies on salts and proteins, 
Canadian Chem. & Process Indust. 34: 204-211, 249, 1950. 

30. Longsworth, L. G.: Interferometry^ in electrophoresis. Anal, Chem. 23: 

346-348, 1951. 

31. Lotmar, W.: Interferometeranordnungen fur Mikro-Elektrophorese, 

Helvet. chim. acta 32: 1847-1850, 1949. 

*32. Luetscher, J. A., Jr.: Biological and Medical Applications of Electro¬ 
phoresis, Physiol. Rev. 27: 621-642, 1947. 

*.33. Macinnes, D. A., and Longsworth, L. G,: in Alexander, J, (ed.): Col¬ 
loid Chemistry (New York: Reinhold Publishing Corp., 1944), Vol 5 
pp.387-411. ■ ’ 

34. May, M., and Felsing, W, A.: The ionization constants of /3 alanine in 

water and isopropyl alcohol-water mixtures, J. Am, Chem Soc 73- 
406-409, 1951. ' ' 

35. Michaelis, L.: Diethylbarbiturate buffer, J. Biol. Chem, 87: 33-35, 









44 . 

Moore, D. H.. The effect of urea on the electrophoretic patterns of 
serum proteins, J. Am. Chem. Soc. 64: 1090-1092, 1942. 

Moore, D H., and White, J. U.: A new compact Tiselius electrophoresis 
^ apparatus. Rev. Scient. Instruments 19: 700-706, 1948. 

Penwton, Q P.; Agar II. D., and McCarthy, J. L. : Ionophoretic analysis 
in agar gels. Anal. Chem. 23: 994, 1951. 

PcTlmann, G. E., and Kaufman, D.: The effect of ionic strength and pro- 

^T!emen\’^'-om’ L. G.: The specific refractive in- 

erment of some purified proteins, J. Am. Chem. Soc. 70: 2719-2724, 

Philpot, J. St L.: Direct photography of ultracentrifuge sedimentation 
curves. Nature, London 141: 283-284 1938 bcaunentation 

cluwtra A self-plotting interferometric opti- 

lU-ine^vtand Fe'ilTiri V 

^ phoresis, Science 108: ;c4-16e°i<l48 

Shcllocslry, T.: in VVeissberger.’A. (ej.): PkyMcal MeMs oj Organic 

1 ()() 















58 . 

Chemialry (New York: Interscience Publisher.s, Inc., 1040), \'ol. 1, 
pt. 2, pp. 1651-83. 

Smith, K. F., and Briggs, D. R.: Electrophoretic analysis of protein 
interact ion: I Interaction of bovine serum albumin and methyl orange, 
J. Phys. A Coll. Chem. 54: 33-47, 1050. 

Stern, K. (1., and Reiner, M.: Electrophoresis in Medicine, Yale J. Biol. 
& Me<l. 10:67-00, 1046. 

Svensson, 11.: Direkte photographische Aufnahme von Elektrophorese- 
Diagrammen, Kolloid Ztschr. 87: 181-186, 1030; 00: 141-156, 1940. 
Svensson, H.: Electrophoresis by the moving boundary method, Arkiv 
Kemi, Mineral. Geol. 22a: 1-156, 1946. 

Svensson, H., and Brattsten, I.: An apparatus for continuous electro¬ 
phoretic separation in flowing liquids. Ark. kemi 1: 401-411, 1949. 
Svensson, H.: An interferometric method for recording the refractive 
index derivative in concentration gradients, Acta chem. scandinav. 
3: 1170-1177, 1949; 4: 390-403 and 1329-1346, 1950. 

Svensson, H.: An experimental technique for micro-conductometric 
analysis of moving boundary systems, J. Am. Chem. Soc. 72: 1074- 

1978,1950. . 

Tiselius, A.; A new apparatus for electrophoretic analysis of colloidal 

mixtures, Tr. Faraday Soc. 33: 524—531, 1937. 

Tiselius, A.: Electrophoresis of serum globulin: II. Electrophoretu- 
analysis of normal and immune sera. Biochemical J. 31. 1464-1477, 

Tiselius, A., and Rabat, E. A.; An electrophoretic study of nnmune sera 
and purified antibody preparations, J. Exper. Med. 69: 119-131, 


Treffers, H. P., and Moore, D. H.: The use of infra-red film for electro¬ 
phoretic and ultracentrifugal analyses, Science 93: 240, 1941. 
Wiedemann, E.: Elektrophorese, Experientia 3: 341-353, 194 <. 
Wiedemann, E.: Neue Zellensatze fiir Elektrophorese-Messungen und 
kleinpraparative Elektrophorese-Versuche, Ilelvet. chim. acU .11. 

Wolt'er, VeiBesserung der abbildenden 

Minimumstrahlkciitizcichmmg, Ann. 1 hys. f. 1S2 1.*-, l.>.*0 


EDWARD G. PICKELS, Specialized Instruments Corporaiion, Belmont, Calif. 

I. Nature of Sedimentation in Centrifugal Fields 

In a gravitational field of force, a particle suspended in a 
liiiuid medium of lower specific gravity tends to migrate through 
the fluid in a downward direction. If the density of the particle 
is less than that of the medium, the movement is, of course, up¬ 
ward. Assuming the absence of convective disturbances and 
electrical forces, the rate of migration is governed by the size, 
shape and density of the particle, density and viscosity of the 
suspending medium and intensity of the gravitational field. In a 
dilute suspension of identical particles, all rates will be equal 
and, for the case of downward sedimentation, those particles 
originally at the surface form a well defined, moving boundary 
which demarcates the supernatant fluid from the sedimenting 
phase. Until the boundary reaches the bottom of the vessel, par¬ 
ticles continue to collect there at a uniform rate and the concentra¬ 
tion of suspended particles in the “plateau” region between the 
boundary and the bottom remains constant. Two groups of par¬ 
ticles of 2 different sedimentation rates in a binary mixture pro¬ 
duce 2 boundaries which become further separated with time, the 
concentration of each component being indicated by the net change 
in concentration across its respective boundarv. Particles of oi-.^led 
size and sedimentation rate exhibit a blurred boundarv "which 
spreads progressively as sedimentation proceeds. 

1 hough well defined, the boundary representing a monodisperse 
system is not infinitely sharp because of the thermal agif,ation or 
hrownia'i movement of the particles. In seneral, the smaller (he 
I arhcles the slower the sedimentation rate and the hisher the rate 
at which the boundary diffuses once it is formed. However a 
jilateau region of uniform concentration continues to exist for a 

els of the fluid which are subject to some backward diffusion 

particles which are increasing in concentration at the bottom of 
he vessel. Sedimentation rates are still measurable siller^ t 

"><■ plateau regional) “'''■‘‘ntr.ation is one-half thai in 




To })ro(hice measurable sedimentation with particles, e.g., 
protein moloe\iles, in the size range considerably below 1 /u within a 
reasonable time and without excessive diffusion at the boundary, 
one must use considerably higher fields of force than afforded by 
gravity alone. I'bani f hough fields in excess of 250,000 times gravity 
are now' routinely used for this purpose, precise measurements ol 
sedimentation rate are still limited to particles having molecular 
weights above about 10,000, because of the distorting effects of 
diffusion on the plateau region and on the boundary before it has 
cleared the meniscus. However, preliminary studies performed 




pjfj I —Concentration distribution of small “globular” protein (s^ou. = 
2S- D-lou, = 10 X 10-L molecular weight = 18,000) at 80 min intei-val.s during 
ultmcentrifugation under average force of 300,000 times gravity. Meniscus 
distance from axis, 5.8 cm. Ordinate: % of original concentration; abscissa: 
distance h-om meniscus, in cm. 

with boundaries beyond the meniscus intlicate the feasibility of 
good measurements with molecular weights down to a,t least 1,000. 
In pauci-disperse preparations, boundary differentiation is limited 
to molecular weights of 10,000 or more, depending on relative 

sedimentation rates, etc. , , • „ 

Through the pioneering work of Svedberg and Ins collaborators 

(35) since 1923 it has been demonstrated that high centrifuga 
forces can be utilized in a practical manner lor producing pie- 
cisely measurable .sedimentation of molecular particles when cei- 
tain‘conditions are met. An essential requirement is that t ie 
containing ves.sel or cell be shaped appropriately lor use in a ladi- 
ally directed centrifugal field of force. It is thus made sectoi- 



shaped, having 2 flat walls which, if extended, would intersect 
nlong the axis of rotation, while the other 2 side walls are perpendic¬ 
ular to tlie axis. Thus, particles originally close to any of these 

walls continue during sedimentation to pursue an average course 
parallel to the wall, so that there is no sedimentation against or 
away from the Avails to introduce convective disturbances. 

As illustrated in Figure 1, there is then 1 primary difference 
l)etAveen sedimentation in centrifugal and gravitational fields. 
Since, during centrifugal sedimentation, the particles on the aver¬ 
age follow diverging radial paths and since their rate of movement 
increases Avith the distance from the axis of rotation, the concentra¬ 
tion in the plateau region steadily decreases. Nevertheless, the 
true boundary position at any time is still A^ery nearly the level at 
AA'hich the concentration is half that of the plateau region at that 
time. Also, there is a definite mathematical relationship (see 
page 119) betAA^een the concentration of the plateau region and the 
displacement of the boundary, so that the initial concentration 
of any differentiated component can still be determined by apply¬ 
ing a correction factor. The curves of Figure 1, taken from an actual 
sample of a small “globular” protein and illustrating the limitations 
imposed by diffusion on the so-called “sedimentation velocity 
method,” are also useful guides in visualizing the behavior of other 
larger, but similarly shaped proteins. It can be shown that to a 

1st approximation, protein boundaries sedimented to corresponding 
positions would require sedimentation times inversely proportional 
to the two-thirds poAA'er of the molecular AA’eight, and the boundary 
spreading Avoidd also be inversely proportional to the tAvo-thirds 
poAver of the molecular Aveight. Thus a boundary of 144,000 molecu¬ 
lar Aveight particles aa^ouIcI reach the illustrated positions at 20 min 
interA^als and Avould exhibit only one-quarter as much spreading. 
Also, at one-half the speed, or one-quarter the average centrifugal 
torce, the same positions Avould haA^e been reached at 80 min inter¬ 
vals and the boundary spreading Avould have been one-half as 
much as Avith the smaller protein, since boundary spreading is 
known to be proportional to the square root of the diffusion time. 

of a monodisperse substance is 
thiX I I'l terms of its “sedimentation constant” s, 

(1 dvne W p«' unit field of force 

S n ^ mass). Sedimentation constants are usually given in 

S or Svedberg units, an S unit being 1 X 10-» cm/sec/dvne nJ 
oice on each gram (equivalent to 1 X 10“^® seci The i.. r 
imentation constants routinely me« b^ analt cll lrf 

1 , ro aoout 1,()00S for some bacteriophages 



which are more than 10,000 times as massive. So that sedimenta¬ 
tion constants may be more directly compared, they are also cus¬ 
tomarily reduced to standard conditions with respect to the vis¬ 
cosity and density of the medium, as discussed under mathema tical 
relationships (page 119). 

A special case of which very practical use is made is that in which 
the sedimentation is kept slow enough in comparison to diffusion to 
prevent formation of any distinguishable boundary. When this 
situation is continued long enough, Avith the centrifugal force and 
temperature held constant, a state of equilibrium is established 
between the diffusion and sedimentation processes. An analysis 
of the concentration distribution then permits the determination 
of molecular weight, in accordance with Svedberg’s “e(iuilibrium 
method” equation (5), given on page 121. Molecular weights of 
less than 100 can be determined (35). Sedimentation in containers 
which are not sector-shaped, or under conditions where freedom 
from convection is not realized, is taken up in section III (p. 126). 

The standard reference work for all phases of ultracentrifugation, 
including theory, method and application, is The i ltracentrifuge 
by Svedberg and Pedersen (35). It also contains an extensive 
bibliography of publications before 1940. Less extensive but more 
recent reviews and bibliographies have been published by Nichols 
and Bailey (19) and Pickels (26). 

11. The Analytical Ultracentrifuge 

The most precise and routinely used methods for measuring 
sedimentation are optical. Basic reciuiremenfs of an optical ultra¬ 
centrifuge are a rotor capable of withstanding high centrihigal 
forces, a transparent cell of sufficient re.solving power (35), 
means for attaining and maintaining high speeds, a method (4 
avoiding temperature gradients and consequent convective dis¬ 
turbances within the fluid sample, means oi measuring and pref¬ 
erably controlling the temperature of the sample, and an optica 
system capable of recording sedimentation Ihe early slov-spju 
machines of Svedberg and Nichols were eleetneally 'I' '™"’ ) 

in later models, steel rotors with integral, rigid 
around a horizontal axis, within an atmosphere ol P ^ 
hydrogen by jets of pressured oil impinging against srnall tin bine 

bfadTat’J shaft ends (35). ^ 

now made to spin in a vacuum around a veitical axis and aie 

llexibly suspended so as to be self-balancing. The is 
comnressed^air (23), mechanical bearings (.20 or, m some (.im. , 
b™ie![!c)ift (1)’’ Driving power is furnished by compressed air 


1 I I 

(2.‘{) or doclric motor (27). Tlie speed and resoh iiif^ j)o\ver of all 
modern ultraeentrifnges are limited only by the strenfiith of the 
i-otoi' and its eell. inexperience has dictated a distance from the axis 
of tii(' rotor to the center of tlie cell of 0.5 cm as being near the 
optimum for general use (35). 





— IS »••• 


. t * 6 8 W 12 r4 

iJist.anc6 From Meniscus 

I rllililiilililifllifli 

Light Absorption Method 
15 Minutes 

Light Absorption Method 
30 Minutes 

i\ 30 Min.0 

Refractive Index 


Herern„« t„ Figure mL Ju,;,:;! ^ and 

.vnhlmped'rquiX u'hich 



cups, 20, for the disks, form a transparent cell. Filling is done with 
a syringe needle through a small hole, which is then sealed with a 
})lastic w'asher and screws phig. Details of construction have been 
described elsew’here (24). The depth of the sectoral cavity is 1.5 cm 
in a radial direction with respect to the rotor, and the thickness be- 
tw^een the quartz disks is 1.2 cm. Capacity is approximately 0.8 ml. 
The cell fits closely into a 1 in. hole provided in the 7V4 in. rotor, 
28, of aluminum alloy, which is capable of routine operation at 
60,000 rpm. The rotor spins within a heavy steel chamber, 27, 
w hich is evacuated to pressures below' 1 ju Ilg b}^ an oil diffusion 

3.—Analytical rotor (7^A i*'- maxiimiin diameter) of 
,-iuab e of routine operation at 60,000 vpm (average force on cell, 260 000 
limi gnwity). Cell counterbalance, cell housing, centerpiece and secto. cup.s 
holding transparent quartz windows are also shown. 

pump, 2.3, and a liacking mechanical vacuum pump U Added 
prutTOtiuu against possible rotor explosion is provideil y a 2 in. 
thick steel guard ring, 16. Opening, closing and automatic locking 
of the vacuum chamber is accomplished through the rotation ot 

q hoavv threaded support rods, 26. 

When’ the rotor is operating at full 'vit i a 

suri oundlngs id .ou t • ^ compressor unit, 

enveloping cyhndti, /o, ^ maintain 

25 'J'he arrangement is esiiecially useliil ttlieii one 



the rotor at reduced temperatures, as in the case of some unusually 
labile materials. Most ultracentrifugal runs at high speed are com¬ 
pleted within about 3 hr, and to determine rotor temperature for 
the computation of fluid viscosity, it is quite sufficient to use the 
mean of values obtained (to within about 0.2° G) with a contact 
thermocouple applied to the rotor immediately before and after 
the run. However, a 2d thermocouple element is fixed in position 
near the coupling device which attaches to the top of the rotor, and 
its meter indicates approximate rotor temperature continuously 
during operation. At full speed, after equilibrium is established 
within about 20 min, the temperature of the fixed thermocouple 
stays about 2° higher than rotor temperature by virtue of the 
frictional resistance of the few remaining air molecules. Once the 
exact difference for a given set of conditions is determined by meas¬ 
urements with the contact thermocouple, the fi.xed thermocouple 
readings may be irsed in corrected form as indicating rotor tempera¬ 
ture. The arrangement is especially useful for long runs. Waugh 
(private communication) has pointed out that theoretically a 
slight lowering of rotor temperature, proportional to centrifugal 
force, should occur during acceleration because of the elastic 
e.xtension ol the metal. For 60,000 rpm, the theoretical correction 

of measurements made on the stationary rotor is about —0.7° C. 
lemperatuie changes of this magnitude have been detected by 

** aiiu oy us wiin "raaiation" thermocouples situated on the 
axis of rotation below the rotor, but our own experiments indicate 
that the teniperature reading is influenced to some extent by the 

residual gas molecules thrown from the rotor 
^ belf-balancmg and thermal isolation of the rotor is afforded by a 
/10 m. flexible steel shaft which passes throue-h a scalino- o-InnH af, 

n cip])i oximately 0.1 per cent 



For detection of sedimentation boundaries {11, Fig. 2) by the 
method detailed in section III, light from a slit source, 21, is paral¬ 
leled bj’’ a collimating lens, 22, and directed through the cell within 
the spinning rotor, 28. In the region of a sedimenting boundary, 
the light rays suffer a deviation which the system is designed to 
measure. The paralleled light is converged by a condensing lens, 
29, to form an image of the slit source at an inclined slit, 2, after 
reflection from a mirror, 1. The light continues through a camera 
lens, 3, and a cylindrical lens, 4, to give on the photographic plate, 
0, and on a viewing screen, 7, following reflection from a partial 
mirror 5, a pattern, 11, from which the concentration distri¬ 
bution, S, of the material within the cell can be determined. Sedi¬ 
menting boundaries, 9 and 10, may also be detected through the 
al)sorption by the sedinuaitating material of specific wavelengths 
of light according to the met hod of Svedberg (35). 

Preparative i-otois, 12, with inclined tube holes of considerable 
capacity may also be spun in the same machine. The filled plastic 
or metal tubes, l.'i, can be individually sealed, 12, before being 
placed in the rotor. 

OPTICAL :metiiods of kecoruing sedimentation 

Cylindrical lens procedure .—The optical method most widely 
used with the ultracentrifuge is that attributed primarily to Phil- 
pot and modified by Svensson, who discussed the method in great 
detail in his thesis (36). A schematic representation of the optical 
principles on which the method is based is given in Figure 4. 
For a more complete description, see reference (26). In the sim¬ 
plest arrangement, all elements of the system are centeied on an 
optical axis and are disposed along it so as to meet the following 
requirements: light rays from a common point m the slit source 
are parallel after passing through the collimating lens; undisturbed 
rays (solid lines) are focused by the condensing lens to form an 
image of the slit source in front of a camera lens as shown; with 
the cylindrical lens removed from the system (as in the 1st 
ment) light through a pinhole in the object plane is focused by the 
camera lens to form a sharp image of the pinhole on the screen 
regardless of any deflection (broken lines) or scattering of the light 
artificially introduced at the pinhole. To complete the system 
(2d arrangement in Fis. 4), a diaphragm with an inclined slit 
is iiiterpoLl at the position of the slit image, and a cylindrical 
ILf«dSraxis vertical is so placed that a diffusely illuminated 
pinhole in the plane of the inclined slit would be focused as a sharp 
Vertical line of light on the screen. It will be *7 

diaphragm with the inclined slit is now removed, light comin„ 



lliiouf>;li a pinhole in the object diaphragm is focused on the screen 
as a horizontal line of light, and its position is the same regardless 
of any vertical deviation (dotted lines) imposed on the light rays at 
the pinhole. With the inclined slit diaphragm in place and inter¬ 
rupting all except 1 of the rays within a light “sheet” (rays in 
single plane) emanating from the pinhole, only a short element 
of this horizontal line of light remains on the screen. This spot of 
light is disj)laced sideward if the inclined slit is moved laterally 

Object plone Imoge of source 

or if the light sheet is deflected a different amount at the pinhole 

so as to change the vertical level at which it reaches the inclined 
slit diaphragm. 

In essence, the system is able to convert a vertical deviation of 
ight at the object plane into a horizontal displacement of a light 
point on a screen or photographic plate without altering the point’s 
vertical height, which itself corresponds to a certain level in the 
object plane. Study of the focusing arrangement shows also that 
the same effect (sharp light point) would be produced on the screen 
e\en if a horizontal slit were substituted for the pinhole in the 
object plane, except that the light point would be brighter on Jhe 


screen. In recording sedimentation boundaries, as illustrated toi 
an ultracentriiuge with horizontal rotational axis in the 3d 
arranpment of Figure 4, a diaphragm with a slit a few millimeters 
^vide is used near the cell to avoid lack of definition which would 
otherwise be caused by curvature of the path followed by any 
element of the fluid around the axis of rotation. Fortunately, we 
are concerned only with the redistribution of material in a radial 
direction, and hence no stroboscopic arrangement to arrest the 
rotational motion of the rotor optically is necessary. When the 
cell is ill the object plane with the middle of a sedimentation 
boundary at B, the light rays passing through the solution at B 
are deviated the greatest amount since this is the region of greatest 
concentration gradient, and hence of greatest prismatic action. 
Rays through the supernatant fluid and through the plateau 
region are undeviated, while those through other portions of the 
boundary are deviated according to the respective refractive index 
gradients. A curi'ed line of light of the general character illustrated 
is thus obtained on the screen, the point of greatest lateral devia¬ 
tion, B', representing the middle of the boundary. 

In the simple examples illustrated (see Fig. 2 also), the curve 
is essentially a graphic plot of the refractive index gradient (directly 
proportional to concentration gradient) for all levels within the 
fluid column. If the effect is integrated from the region just above 
the boundary to the plateau region just below it, one obtains the 
total refracti\'e index change associated with the boundary and 
hence the concentration of the material which constitutes the 
boundary, assuming that the refractive properties of the material 
are known. In other words, the concentration of any resolved 
component may be determined by measuring the area under the 
respective peak (down to the base line corresponding to zero re¬ 
fractive gradient) of the ivfractive index gradient curve, and 
applying a formula inv'olving the j)h 3 '^sical constants of the system 
(p. 119). For measurement of sedimentation rates, the system 
is usually ideal, since the boundary position for a dilute homo¬ 
geneous material is given by the tip of the peak. 

In actual practice, the simple method outlined for determining 
the concentration of a component would apply accurately only at 
very low speeds of rotation. In high fields of force there aie ex¬ 
traneous deviations of light due pi'irnarily to the hydro.static com¬ 
pression gradient set up in the fluid and secondarily to the minute 
displacement, distortion and compi-ession of the (luartz windows 
within the cell. The partial sedimentation of salts or other low- 
molecular Aveight materials which diffuse rapidly enough to ino- 
duce gradients throughout the fluid also has an effect. For approxi- 



mate area measurements, the displaced base line pertinent to a 
particular boundary can usually be interpolated from the flat or 
gently curving sections of the pattern on either side of the bound¬ 
ary. But for the most accurate measurement, one must obtain a 
“reference base line” by making a duplicate run with the suspend¬ 
ing medium alone, the photograph being taken at the same stage 
of the run. The difference in the patterns, obtained by matching 
the undeviated references, represents the effect due to the partic¬ 
ular component or components not included in the suspending 
medium. In comparing photographs (especially when only 1 hole is 
used in the cell counterbalance) and in accurately measuring 
boundary positions, it is desirable to have the meniscus and the 
radial level in the rotor corresponding to the counterbalance hole 
well defined across the whole photograph. This is not accomplished 
when an inclined slit (Fig. 4) is used. 

Instead of having a light pattern on a dark background, it is 
best to use a dark pattern on a light background. Hence, an in¬ 
clined opaque bar about Vie in. wide or a straight wire about 
0.010 in. in diameter is usually employed instead of an inclined 
slit. In measuring areas from photographs obtained with the wire, 
one should follow the middle of the pattern, which has a minimal 
apparent width and sharpness regardless of wire size because of 
optical diffraction. Also by reason of diffraction, the apparent width 
of the pattern (measured parallel to the meniscus line in the photo¬ 
graph) obtained with the bar changes Avith slope of the curve and 
with optical density of the photographic negative. To avoid errors 
which might amount to 15-20 per cent in the case of very light 
photographs, one should trace both edges of the pattern and use 
the mean. In general, the wire is preferred for simple patterns 
because of easier tracing, but in the case of sharp multiple bounda¬ 
ries close together, the bar pattern usually affords better resolution. 

I have also been able to demonstrate that patterns for 2 separate 
cells lun simultaneously in the same rotor can be recorded in usable 
form in a single photograph. The usual counterbalance is replaced 
by a “prismatic” cell which is the same as the standard cell except 
tliat the outside surface of the upper quartz disk and the corre¬ 
sponding surface of the sector cup are sloped (but still flat) at about 
1 to a hue which perpendicularly intersects the axis of rotation 
and the axis of the cell. This gives a separation of the 2 patterns, 
borne loss m optical contrast and a limitation on the amount of lio-ht 
deviation which can be ivconled must be t olerated. However a He- 
memlous saving in lime is alIor(l,.,l u lien a laige miinliei- of samples 
must he lun, and routine use lias been made of tlie sclieme witli 
good success ill certain studies (<)). Although light through the i 







refraction rat tern of fluid 

component of higher molecular weight 

Fig. 5.—Serial photographs illustrating use of the Philpot-Svensson optical 
system, with inclined bar instead of slit, to photograph sedimentation in a 1 
per cent solution (medium of 0.2M Na(d) of gamma globulin {s 20 w = 7.2S 
for main comj)onent) at 59,780 rpm. faglit to form unrefracted references 
comes through holes in the cell counteiLalance. 

g—P hotographs of refractive patterns obtained with an inclined wire 
(ahooe) and an inclined bar {behw). That at lower rift, showing 2 relerences 
rom cell counterbalance, is of seruni albumin 
for very small amount of more rapidly settling 

iLw patterns obtained when both a normal f,, b' at 52 oto r 
simultaneously, with water at low speeil (upper left), water at n> 

(upper right), and 1.5M NaCl after 15 min at same speed (flower left). 

bubble above the fluid can be used as an undeviated reference, 
a more dependable and better defined one is obtained by dulling 
a small vertical hole through the edge of the rotor in a positioi 
at'right angles to the 2 cells. The various points discussed above 

are illustrated in Figures 5 and 6 ^ 

Modified scale method. -Anoihev optical method xnIucIi 


the same sort of information as the one described is the scale 
method of Lamm, later simplified by Svensson (37). Scale lines 
are photof^raphed through the cell and the displacements of the 
liiu; images are measured and graphically })lotted to give the 
refractive index gi-adient curve. The method is tedious, although 
it is least susceptible to the small optical errors that might be 
significant if one wishes to determine refractive index gradients 
with the highest possible precision. It is most ideally suited for 
s(Hlimentation equilibrium studies but is hardly justified in the 
measurement of sedimentation rates. 

Absorption method .—Boundaries may also be detected by the 
absorption method of Svedberg (35), but complete analysis of the 
photographs is not nearly so direct or absolute in measurement 
and the method is seldom used except for certain special appli¬ 
cations. Ultraviolet light is usually employed, and in the case of 
nucleoproteins and certain other materials which absorb ultra¬ 
violet light very strongly, some advantage is gained in the way 
of extending measurements to very low concentrations or un¬ 
usually inhomogeneous material or of differentiating highly absorb¬ 
ing materials in a mixture. 


The derivations of equations for sedimentation constant, molec¬ 
ular weight, frictional coefficient, and concentration correction 
factor have been given by Svedberg and Pedersen (35), and these 
should be consulted for a complete understanding of ultracentrif¬ 
ugal theory. Only the more important of these equations are 
lestated here, along with some less quoted relationships which I 
have found useful. The movement of a sedimentation boundary in 

a (entrilugal field is described in terms of a sedimentation con¬ 
stant, s 

(.log X2 — log g;,) 

< 0^(4 


( 1 ) 

where CO is the angular velocity in radians/sec ( =27r X rps) 
and X IS the distance in centimeters of the boundary from the axis 
of I otation at time t. A sedimentation constant is usually expressed 
m S or Svedberg units and, when not otherwise indicated it is 
geneially assumed to represent the value obtained by extrapolating 
experimental results to zero concentration, since sedimentation 
rate changes with concentration. Also, for purposes of comparison 
dimentation constants are generally reduced to values which’ 
theoretically would be obtained if the medium Ivid tha 

and density of water at 20 CCsuteeript 2 !) ' 


« ’? 20«(1 - 1 » 

Avheie rj is the viscosity ol tlie niediuin, p is tlie doiisily of the 
medium, and V is the partial specific volume ol the sedimenting 
material. This corresponds to the increase in \'olume when 1 g of 
dry solute is added to a large volume of the medium and is equi\'a- 
lent to the reciprocal of the density of the material (in the un¬ 
hydrated, unsolvated state unless otherwise designated). This 
equation assumes that V would have the same value in the stand¬ 
ard medium, and it also assumes, when V is taken for the unsol- 
\’ated state, that the sedimenting particles are solvated with 
material of the same density composition as the surrounding 
medium and to the same extent volumetrically in both mediums. 
Usually, the experimental medium and hypothetical standard 
are similar and the densities of solute and medium are not too 
close together. The equation can then be applied with some assur¬ 
ance. When these conditions are not realized, appreciable error 
can be introduced unless proper correcting factors are applied 
(31). y has a value near 0.74 in the case of many proteins. 

Molecular weight, M, can be computed from Svedberg’s formula, 
which assumes that the particles are solvated, if at all, with ma¬ 
terial having the same density as the medium 

M = 


D(l - Up) 


R is the gas constant (8.313 X 10^), T the absolute temperature, 
and D the diffusion constant which is usually measured in a sepa¬ 
rate experiment by standard techniques (20). Values of s and D 
must be for very dilute concentrations (preferably extrapolated to 
zero concentration) and should ideally be for the same medium at 
the same temperature. However, an equation analogous to (2) can 
usually be applied with negligible error to correct s values to the 
conditions of the diffusion experiment. 

The diameter, d, of spherical particles (applies strictly for un¬ 
solvated particles unless V is for solvated material) or a minimal 
possible value for the average diameter of particles of unknown 
shape is given, in the case of dilute solutions, by 


■" (1 - Up) 


For sedimentation equilibrium in a monodisperse system, the 
Svedberg formula for molecular weight is 



(I -l»a,*(.ro=-.r,=) 

where C\ and Co are coiieeiitrations at radial distances, Xi and 
Better suited for nse with the refracti\'e index system described is 
tlie form 

2RT. log [x, (f^)] 

(1 - Fp)co2(a-22 - a-F) 

In equations (5) and (6), the same assumption is made regarding 
solvation as in equation (3). 

The relation between the concentration of a particular compo¬ 
nent in its respecti\'e plateau region at any time t and the original 
concentration Co is 

C. = C. (I)’ (7, 

where Zo and Zt are the respective radial distances of the boundary 
at zero time (meniscus or bottom of cell) and at time t. 

From patterns obtained with a Philpot-Svensson optical system 
using paralleled light through the cell, the concentrations of sedi¬ 
menting components can be computed by applying the following 
equation ^ 

Co = 


A. (• tan 6 

mi -nii-H • An 


( 8 ) 

e IS the angle between the uncleviated slit image and the inclined 
wire, bar or slit; L is the distance in centimeters along the optical 
path from the nodal point of the condensing lens (midplaiie of 

of ceh Z'irin” fif inclined wire; m, is the magnification 

of the ^ ^ “ i‘t‘’ Pliotographic plate, and m, the magnification 
of the cylindrical lens system, determined by photographing scale 
fines which arc placed in the same plane as the inched Tire “ d 
paralle to it with e = 90«. H is the thickness of the fluid column 

tive incremenrof'’tl e i** ‘he specific refrac- 

live mciement of the particular component feoual fn onmco • 

he case of bovine serum albumin, for exam^k^X ydloffight‘i^ 

m.ade if the area is apporti^d L determinations can still be 
to tfie meniscus, l™rttXXVte 



approach on the curve of the sedimenting material. A better method 
is that of drawing a gaiissian curve for each component in such a 
manner that the several curves account for the total concentration. 

The Pei-formauce Index, Pi, is a rating projjosed by the author 
to be applied to preparative rotors, with inclined tubes or with 
fluid cavdties of any shape, to express approximately their relative 
pei’formances in accomplishing the complete precipitation of any 
given material under certain idealized conditions. It depends only 
on the physical dimensions of the rotor and its speed. 

1<'K H-i — log Hi 


where R-i and Ri refer to the maximal and minimal l adial distances 
(from axis of rotation) in centimeters of the contained fluid. 


Alost macromolecules are electi ically charged with respect to the 
suspending medium, and in sedimenting they attempt to drag 
along through the medium those surrounding ions for which there 
is an electrical attraction. Sedimentation rates may be ieduced by 
more than 20 per cent in some cases unless this Donnan effect is 
repressed through the addition of low molecular electrolyte to the 
medium. For most proteins, the reduction in rate is less than 1 per 
cent if 0.2 mole of sodium or potassium chloride is added for each 
1 per cent of protein (35). In the study of small molecules by the 
equilibrium method, equations allowing for charge effect must be 
used (35) Also, it should be emphasized that in citing ultracen¬ 
trifugal results, one should fully specify the composition of the test 
sample. Concentration of the solute, pH, temperature, concentra¬ 
tion of dissolved salts or other materials and the previous history 
of the sample may all have some effect on the corrected sedimenta¬ 
tion constant or the degree of ultracentrifugal homogeneity. 

Sedimentation constants should be measured at several diffeient 
concentrations and the results extrapolated to zero concentration 
for assigning values applicable to the equations given earlier. In 
the case^of “globular” proteins, the sedimentation rate is generally 
decre^ed in the neighborhood of 8 per cent for each 1 per cent o 
nrotein The effect increases with asymmetry m particle shape. If 
U't: Ire limited to a preparation of fixed compos.t.on .g un¬ 
diluted blood serum, or if for any reason s ™lues har e not be 
pvfranolated to zero concentration, then the viscosity of the com 
"ep^ a,Hi approximate correction (J.'- 

!:“':;':’a:iSnsl2) ird; ; oiblah. fair ..true values. 



A discussion of the effects of concentration and polydispersity has 
been presented by Kinell (13). Wlien concentrations are above 1 or 
2 per cent boundary sharpening effects may complicate the analy¬ 
sis. Also, with mixtures in which the sedimenting components con¬ 
tribute appreciably to the viscosity of the solution, sedimentation 
behavior will in general depart, from that observed at infinite dilu¬ 
tion. John.ston and Ogston (12a) have studied this phenomenon. 
The influence of hydrostatic compression on density and viscosity 
may generally be disregaided with aqueous solutions but must lx* 
corrected foi- in the case of organic solvents (26). 

Sedimentation rates are best determined by taking photographs 
at inter\'als and graphically plotting log x against t. The equivaleni 
starting time (t = 0, and radial displacement of meniscus or l)Ot- 
tom of cell for Xo) is time at which approximately two thirds of 
operating speed is reached, assuming nearly uniform acceleration. 
This point, which can usually be employed with small error despite 
reflected diffusion while boundary forms, is valuable where bound¬ 
aries of a mixture do not resolve until late in the run. For example, 
with the ultracentrifuge illustrated, resolution is barely obtained 
with equally concentrated “globular” proteins of 18,000 and 40,000 
molecular weight. For best results with the optical system illus¬ 
trated, the illuminating slit width should be limited to about Vio 
mm, approximately monochromatic light should be used, and the 
photographic exposure time should not be more than about 2 per 
cent of the time required for the boundary to tra\^erse the cell. 
When it is not practical to determine the diffusion constant in a 
regular diffusion apparatus, an approximate \'alue can be obtained 
iiom the boundary in the ultracentrifuge, especially if measure¬ 
ments are made after the rotor is brought to a very low speed fol- 
lovnng a rapid migration of the boundary to the middle of the cell 
at high speed (26). Monodisperse boundaries are sharpened while 
m motion because of the variation of s ^vith concentration; on the 
other hand, mhomogeneity in rate during sedimentation mav give 
the appearance of unduly rapid diffusion. The first and photo- 
gi aphs taken at speed should always be compared to make sure 
that no leakage of fluid has occurred. 

Accurate measurement of concentration from refractive patterns 

in t^ToL? Fm thf"" I“d its proper alinement 

cell narte sho^^l he ? r T ® line run, all 

< ell parts should be identically positioned and oriented only the 

sealing washers should be changed, the cell should be tidi’tened and 

If r"-' '>*• 



tor tlie innor half of tlio cell; cell bottom positions should Ix^ 
matclx'd for the outer half of the cell. Small “spikes” or “ragged¬ 
ness” in the pattern generally denote localized convections which 
could be, l)iit seldom are, due to leakage or poor \ acuum. Their 
appearance at constant speed may indicate rough, incorrectly 
shaped or improperly oriented walls in the centerpiece; or they may 
simply denote local instabilities in the density distribution set up 
within the fluid by the sedimentation process. Especially when a 
high concentration of a highly diffusible salt is incorporated in the 
medium (see Fig. 6), an appreciable redistribution of density is 
accomplished within a short time. The slight reduction in sedi¬ 
mentation or “flotation” rate (9) of protein particles (in their own 
plateau region) as they move into the plateau region of the salt 
where their rate should be uniform can result in a “piling up” 
which tends to produce a layer of fluid that is more dense than the 
adjacent layer on which it is supposed to float. A “hump” which 
always appears in the refractive pattern at the same level of the 
cell, with solute but not with the medium alone, suggests lack of 

flatness in sector walls. 

In starting a run, it is good practice to take 1 photograph at low 
speed (about 4,000 rpm) and another as soon as operating speed is 
reached following rapid acceleration. Any lailure of the refractive 
pattern (excepting regions near meniscus and cell bottom) in the 
1st full speed photograph to match the corresponding reference 
base line from the solvent run is usually caused by the settling of 
very large particles. Differences in optical density and sharpness of 
pattern between the low and high speed pictures is substantiating 
evidence of rapidly settling particles, e.g., denatured protein. 
Appreciable curvature in the low speed pattern, or failure to match 
the corresponding pattern obtained with solvent only, indicates an 
unclean or incompletely dried cell centerpiece, a lack of complete 
mixing in the sample, or faulty assembly of the cell. 

Whereas sedimentation velocity runs ne^'er require more than a 
few hours with aqueous mediums, sedimentation equilibrium runs 
require considerable time and, except for materuxls of very low 
molecular weight, relatively low speed (see (35) for details). loi 
example, a monodisperse “globular” protein of 40,000 molecular 
weight would require about 4 days of steady running at appro- 
priate speed of about 11,000 rpm. It will be noted that equation (6) 
can be applied for a monodisperse protein, even if the lefracti 
increment of the material and the optical constants of ms“-u- 
ment are unknown. If polydisperse materials are to studi^ V 
the equilibrium method, one needs such in ormat.on and should 
apply it to other formulas as outlined by Wales el al. (39). 



The sej)aration cell of Tiselius, Pedersen a^id Svedberg (35), 
though not an ideal solution to the problem, does offer some help 
in correlating sedimentation boundaries with biologic or chemical 
properties. Two liah^es of the cell, which can be individually 
sampled, are separated by a porous partition. Whether or not a 
boundary has crossed the partition before stopping of the ulti-a- 
centrifuge can be observed with the optical system. 


With careful techniques and a stable material, one can repeat 
determinations of sedimentation constant to within less than 1 per 
cent or to within a few hundredths of a Svedberg unit in the case 
of low s values. Kegeles and Gutter (12b) have been able to repro¬ 
duce values for proteins to within less than 0.01 Svedberg unit by 
determining boundary positions from the Fresnel diffraction pat¬ 
terns which occur in the photograph at the boundary when an in¬ 
clined bar is used with the refractive method already described. 
However, aA^erage deviations for ^'alues appearing in the literature 
are sometimes several times as great, probably most often owing to 
the unstable nature of the material or to small errors in the meas¬ 
urement of rotor temperature. An error of 1 ° C usually causes an 
error in the s value of more than 2 per cent. With monodisperse 
systems of large particles, concentrations as low as 0.01 per cent 
can be easily detected and measured. Ten times as much material 
may be required for low molecular weight materials. In the measur¬ 
ing of protein concentration, for example, results can be reproduced 
(at least in the same ultracentrifuge) with a mean deviation of 

about 0.01-0.03 per cent, depending on the sharpness of the bound¬ 

As a general rule, when the refracti\'c pattern of a boundary is 
symmetrical arouml the point of greatest refractive index gradient 
the material constituting the boundary is very nearly homogeneous 
with respect to sedimentation rate. Of course, comparisons can be 
made with results from diffusion experiments or between bound- 
anes brouglit to the same level in the cell at different rotor speeds. 
Ivinell (1.3) and others (35) hat'e treated the subject of polvdisperse 
materials. It should also be noted that aggregates of 2 or 3 primary 
jjarticles may produce discrete secondary boundaries. In addition 
to determining sedimentation rate, molecular weight, homogeneity 
and concentration one may also gain some information regarding 
part cle shape if the approximate degree of solvation is known or 
an be determined as by the method of Sihtola and Svedberg (32) 
If, for example (21), a protein of known molecular weightls hv 
drated with 20 per cent water (not significantly differenUii deLt; 


from medium generally), the partial specific volume and the 
diameter (assuming spherical shape) of the hydrated particles can 
be computed and applied to equation (4), from which a theoretical 
s value for spherical particles can be obtained. The ratio of this to 
the experimentally observed value gives a “frictional ratio” from 
which an “axial ratio” for the particles can be computed. 


For a partial list of publications dealing with applications of the 
ultracentrifuge, one may consult bibliographies prepared by Sved- 
berg et al. (35) and by others (19, 26). It is impossible to summarize 
or to do justice in a limited space to the thousands of reports on 
the ultracentrifugation of materials of many sorts. To give some 
idea of the versatility of the method as applied to substances of 
interest in the field of medicine, I have included in the present 
bibliography some publications selected at random with no at¬ 
tempt to pass on their relative importance. The studies on plasma 
or serum proteins (2, 3, 18, 21, 34, 35) have been particularly fruit¬ 
ful. It has been well established that these proteins have molecular 
weights ranging from about 69,000 in the case of albumin to 
several hundred thousand for some of the globulins, and even to 
several millions in the case of some respiratory proteins from the 
invertebrates. Most plasma proteins have been found to have axial 
ratios ranging from about 1 to 6, but, not too surprisingly, fibrino¬ 
gen appears to have a ratio of almost 20 (21). Recent work of par¬ 
ticular interest is that of Gofman et al (8, 9) on the lipoproteins as 
related to atherosclerosis. In their work, the lipoprotein fractions 
are separated from the other serum proteins by a flotation process 
involving preparative ultracentrifugation after the addition of con¬ 
siderable salt to the serum. In the analytical ultracentrifugation of 
these fractions, the boundaries also move toward the axis of rota¬ 
tion through the salt medium. 

Studies have also been made of whey proteins (5), ovomucoid 
(6) influenza virus (7), visual purple (10), mitochondria (11), 
bacteriophage (15), tobacco mosaic ^drus (16), hornmnes 17 , 
pepsin-treated globulin (22), hemocyamn as affected by Xrays (29), 
ferritin (30) and chicken feather keratin (40). 

III. The Preparative Ultracentrifuge 

Although the ultracentrifuge illustrated in Figure 2 can be used 
for preparative purposes, simpler machines, such as that illustrate 

in Figures 7-9 are generally used. , u i 

Material to be centrifuged is placed in test-tube shaped o - 


tainers of plastic or metal which can be individually sealed with 
special caps having a rubber washer that expands against the con¬ 
tainer wall when the cap assembly is tightened (Fig. 7). Weighing 
of the tubes for balance is not necessary because of the self-balanc¬ 
ing features of the machine. Sealed tubes are placed into holes of 
an aluminum alloy rotor, 4, with enough water first placed in the 

Fi(i. 7 -ScJ.oiUHti.r .Iniwiiifr of a comiuoiriallv |)n,(lu,-,o(l, cloc;!ricallv drivon 

e mecha" pump, 17. As in the case of the analytical macLe 
there aie a suard cylinder, B, and a liner, 2, which is refrigerated hv 
a compressor unit. The electric drive 1 is sih„t»,l tli *i 
vacuum chamber, and a V„ in. steel sliafl from it if fitted It its 
ippei end with a spud on which the rotor rests. Extra cooling is 



fiiriiislied by a blower, 18. Details of the drive and other parts of 
the machine have been described (28). 

Through a manual control, 7, the steel lid, 8, of the \'acuuin 
chamber can be drawn into position and allowed to drop onto a 
large vacuum-sealing gasket. Another control (not shown) governs 
opening and closure of the air valve, 15, and starting of the vacuum 
pump. A time clock, 9, controls the length of the run, and a speed 
adjustment, 13, provides preselection of any speed up to 40,000 
rpm. Switches, 10, 12, beside the speed indicator, 11, control the 

Fig 8 —Interchangeable rotors of aluminum alloy and flexible plastic tubes 
usJd whh tifpreparltive ultracentrifuge (Dg. 7)-f-^est roto^^ 
tpr 160 ml canacity) has routine speed rating of 40,000 ipm, the largest t 
h! ’.iLmete^atacfty approximately 1 I) is rated at 20.000 rpm. (Courtesy ot 
Specialized Instruments Corp.) 

power supply and refrigerator and allow tor preselection <>f» 
rate. After the rotor, which carries its own handle (hig. S), is 
place and the chamber has been evacuated for 1 or 2 mm, die ma¬ 
rine is started with a jiiish button, H, and goes through the com¬ 
plete cycle of operation automatically, controlled by an electionic 
unit S A 7 in.. 12-tiibe rotor of 160 ml capacity acce crates 
smoothly up to the maximum spi'cd of 40,000 rpm (144,00 imes 
’iiTavitv) in 4 min, is held at constant speed lor the selected tune, 

and is'tiien decelerated in a 

speed from which it then very grirdnally drilts to lest. lemperatnie 



rise, even at top speed and without refrigeration, amounts to only 
a few degrees centigrade per hour for runs at room temperature. 
The tubes range in diameter from V 2 hi. (2.5-3.5 in. long) to IV 2 
in. (4-5 in. long). A 1 1 rotor with inclined tubes requires about 15 
min for acceleration to its limiting routine speed of 20,000 rpm. 
Tube angles (to axis of rotation) ranging from 14 to 40° are used. 
Each rotor is provided Avith a “centrifugal fuse” which, ivhen acti¬ 
vated by accidental overspeeding of the rotor appreciably abov^e 
its rating, will interrupt the circuit supplying power to the drive. 

The nature of sedimentation in inclined-tube rotors has been 
described (25). In general, precipitate is initially collected at the 
bottom of the tube at a more rapid rate than if the same fluid were 
contained in a sectoral cell with the inner and outer edges of the 
fluid at the same radial distances from the axis of rotation. How¬ 
ever, assuming the absence of any convective disturbances except¬ 
ing that caused by the centrifugal concentration of sedimenting par¬ 
ticles against the side walls of the tube, the last particles to reach the 
bottom of the tube require approximately the same total time as if 
they had moved between the same radial distances in a sectoral cell. 
Hence the approximate time to produce a nearly complete precipita¬ 
tion ol a material of known sedimentation constant can be pre¬ 
dicted (if due allowance is made for diffusion, if significant), and 
equation (9) would appear to afford a fair basis on which to com¬ 

pare various rotor designs. 

As a rnatter of fact, under appropilate conditions sedimentation 
boundaries can actually be detected and rates measured by ex¬ 
amination of tubes with partially precipitated material or by a 
sampling of the fluid column and a testing of the aliquots for some 
specific property (25, 7, 4). In such studies, best success is had by 
using tubes of small diameter (V 2 in.), an appreciably large tube 
angle (35-45° preferably), high concentrations of material (1 per 
cent or more) and a pod vacuum. Because of thermally induced 
convection results with rotors spinning in the open air are generallv 
poor; in the case of vacuum ultracentrifuges, negligible increase in 
efficiency is accomplished by the ailditioii of an oil diffusion pump 
except in cases of unusually difficult separations. The iiicoi-poratioii 

gi-atlient of sucrose or other material in the 
solution as a stabilizing agent is necessary in the case of very low 
concentrations (25, 7). 

venMair'vol'''■’> ‘-‘’"v-nient for handling 
UIJ laige lolunies ol fluid, especially with materials of relatii elv 

gh sedinientation rate. 'Phe amount of fluid must be such th'^ 
dining operation It almost touches a solid removable core in order 
.0 prevent slo.shing anil the development of™ uiZlait 



during sUirtiiig and stopping. Sediment may be resuspended from 
the grooves at the periphery by turning the rotor witli its axis hori¬ 
zontal while stainless steel balls are allowed to roll around inside. 

It should be recognized that in large tubes and in single-cavity 
rotors, a little restirring of sediment from the top, uncompressed 
layer of the sediment might be expected during deceleration, since 
the fluid tends to keep up the rotation. Also, in batch-type or 
single-cavity rotors, and to a lesser extent in very large tubes (de¬ 
pending on concentration of material), the minor thermal gi’adients 
existing even with the rotor spinning in a fair vacuum, and to a 

p.j; q —Hollowetl-out hatoli-typo rotor for the i)rei)iirative ultnicentrifup! 
(FiK 7). Oiaineter is 9 in., rapacity 1.7 1, speed 20,000 rpm C.apacity can l)e 
alnfost doubled in larger, slower rotors. (Courtesy ol Specialized Instruineiits 


lesser extent the minor fluctuations in speed, may cause the sedi¬ 
mentation behavior to approach what might be called ‘‘asymptotic 
sedimentation.” This refers to a situation m which only the layer 
of fluid immediately adjacent to the wall or packed sediment re¬ 
mains stagnant. The main body of the fluid is constan ly, though 
perhaps slo^^•ly, circulating and remixing so that the total amount of 
materkil being deposited in the stagnant zone per unit time is 
dii^cily propm-tional to the concentration in the mam body o the 
fluid Thus, if a certain eflective centrilugation time piecipi a.^_ 

half of tl,e ;sfarting material, doubling the '™;; ‘Vrsrv'Ve" 
per cent separation and tripling the time would lemore 8, /, pei 


(•(' 111 ; l)ul tiu'orc'tically it could never be completely precipitated. 
However, computations for typical rotor designs and preliminary 
experiments have shown that ~Iz~^Ia of the material is removed in 
the time theoretically required for complete precipitation in the 
absence of convection, and hence with v^acuum centrifuges equation 
(9) still furnishes a fair method of comparing rotor designs, and 
approximate precipitation times for known materials can still be 

The Sharpies continuous-flow supercentrifuge has been found 
useful in the concentration of some materials, particularly the 
larger viruses, and sevei-al articles (33, 38) describing its applica¬ 
tion have been published. 

Notk. —This section was reviewed hy John Tiawrence Oncley and bona 
A. Lewis. 

8 . 

9 . 

10 . 


12 . 






ilcanis, .1. W.; ll<)s.s, .1. J)., and Dillon, J. F.: MaKiielicallv suspended vac- 
uuin-tvpc ultracentrifuge, Kev. Scient. Instruments 22: 77, February, 

Blix, Ciuiinar, and Pedersen, K. O.: Electrophoresis and ultracentrifuga¬ 
tion of lipid-free human serum, Acta chem. scandinav. 1:511, 1947, 

Cann, J, R., et al.: Ultracentrifugal studies of gamma globulins prepared by 
electrophoresis-convection. Science 114:30, 1951. 

Curnen, E. C.; Pickels, E. G., and Horsfall, F. L.: Centrifugation studies on 
pneumonia virus of mice (PVM), J. Exper. Med. 85: 23, January, 1947. 

Deut.sch, H. F.: A study of whey pjoteins from the milk of various animals, 
J. Biol. Chem. 169:437, July, 1947. 

Fredericq, E., and Deut.sch, H. F.: Studies on ovomucoid, J. Biol. Chem. 
181:499, December, 1949.' 

Friedwald, W. F., a,nd Pickels, E. G.: Centrifugation and ultrafiltration 
studies on allantoic fluid preparations of influenza virus, J. Exper Med 
79:301, March, 1944. 

Gofman, J. W.; Lindgren, F. T., and Elliott, II.: Ultracentrifugal studies 
of lipoproteins of human serum, J. Biol. Chem. 179:973, June, 1949. 

G(^ma,n, Lipoproteins and atherosclerosis, J. Gerontol. 6: 105, 

Hecht, S., and Pickels, E. G.: The sedimentation constant of visual pur¬ 
ple, I roc. ^lat. Acad. Sc. 24: 172, April, 1938 

HogcLoom, G. ir and Schneider, W. 0.: Proteins of liver and hepatoma 
mitochondria. Science 113: 355, 1951. * 

ilolmberg, C. G.: Studies on the splitting of fibrin under the influence of 
17ATjaImary,T94r'“'’^^^^ streptococci, Arkiv Kemi, Mineral. Geol. 

. Johnston J. P., and Ogston, A. G.: Tr. Faraday Soc. 42: 789, 1946 

. Kegeles G., and Gutter, F. J.: The determination of .sedimentation con 
1951 P^itteriis, J. Am. Chem. Soc. 73: No. 3770, 

Kinell, P.: On the determination of iiolvdisixM-sitv fmm ,.u.. * 'i- 

ugal sediment at ion, Acta chem. scandinav. 1- 33 V-350 1947 ' 

Kmell, P.: tiueh.ues re,narr,ue» eur PWet doia eolfe’nlrMion et de 



la polydispersion sur la sedimentation par ultracentrifugation, J. cliini. 
phys. 44; 4, 1947. 

15. Kozloff, L. M., and Putnam, F. W.: Biochemical studies of virus reproduc¬ 

tion, J. Biol. Chem. 181: 207, November, 1949. 

16. Lauffer, M. A.: The size and shape of tobacco mosaic virus particles, J. 

Am. Chem. Soc. 66:1188, 1944. 

17. Li, C. H.; Simpson, M. E., and Evans, H. M.; Isolation of pituitary follicle- 

stimulating hormone. Science 109:445,1949. 

18. Lindgren, F. T., et al.: The ultracentrifugal composition of normal rabbit 

serum, J. Biol. Chem. 182; 1, January, 1950. 

19. Nichols, J. B., and Bailey, E. D.; Determinations with the Ultracentrifuge, 

in Weissberger, A. (ed.): Physical Methods of Organic Chemistry (2d ed.; 
New York: Interscience Publishers, Inc., 1949). 

20. Neurath, II.; The investigation of proteins by diffusion measurements, 

Chem. Rev. 30:357, June, 1942. 

21. Oncley, J. L.; Scatchard, G., and Brown, A.; Physical-chemical character¬ 

istics of certain of the proteins of normal human plasma, J. Phys. & Coll. 

Chem. 51:184, January, 1947. 

22. Petermann, M. L.; Ultracentrifugal analysis of pepsin-treated serum globu¬ 

lins, J. Phys. Chem. 46:183, January, 1942. 

23. Pickels, E. G.; A new type of air bearing for air-driven high speed cen¬ 

trifuges, Rev. Scient. Instruments 9: 358, November, 1938. 

24. Pickels, E. G.: Ultracentrifuge cell. Rev. Scient. Instruments 13; 426, Oc¬ 

tober, 1942. 

25. Pickels, E. G.: Sedimentation in the angle centrifuge, J. Gen. Physiol. 26: 

341, January, 1943. 

26. Pickels, E. G.': Centrifugation, in Uber, F. M. (ed.): Biophysical Research 

Methods (New York: Interscience Publishers, Inc., 1950). 

27. Pickels, E. G.; Precision speed control. Machine Design 22: 102, Septem- 



ber, 1950. ♦ 

Pickels, E. G., and Scofield, P. F.; Model L preparative centrifuge. Elect. 

Mfg.45:66, January, 1950. . . 

Pickels, E. G., and Anderson, R. S.: Molecular association of hemocyanm 
produced by x-rays as observed in the ultracentrifuge, J. Gen. Physiol. 

30: 83, November, 1946. , i nv. 

30. Rothen, A.: Ferritin and apoferritin in the ultracentrifuge, J. Biol. Chem. 

152:679, March, 1944. . • r j- * 

Schachman, H. K., and Lauffer, M. A.; The density correction of sedimenta¬ 
tion constants, J. Am. Chem. Soc. 72: 4266, September, 19o0 
Sihtola, H., and Svedberg, T.: A new method to determine the degree of 
solvation of macromolecules, Acta chem. scaiidinav. 2: 4/4, 1948. 
Stanley W. M.: The efficiency of different Sharpies centrifuge boiiIs m 
the concentration of tobacco mosaic and influenza viruses, J. Immunol. 




34. Svedblrl’ t-d utieniu-s A.: The .edime,.Ution co„.ta„.. of the 

atory proteins, Biol. Bull. 66: 191, April, . • v^ri-• Ovfnnl 

35 SvedbJrg, T., and Pedersen, K. O.: The Ultracentnfvge (New \ ork. Oxfoid 
30. i..v the n,ovh,« houn.hoy n.ethod, Aehiv 

37. SveiLToii!^n!rAsh(..t-f().-u^ 



39. Wales, M.; Williams, J. W.; Thompson, J. O., and Ewart, R. H.: Sedimenta¬ 

tion equilibria of polydisperse non-ideal solutes, J. Phys. and Coll. 
Chem. 52, No. 6; 983, June, 1948. 

40. Ward, W. II.; High, L. M., and Lundgren, H. P.: Physicochemical charac¬ 

terization of dispersed chicken feather keratin, J. Pol. Res. 1, No. 1: 22, 

41. Williams, J. W., and Baldwin, R. L.: J. Am. Chem. Soc., 72, 4325, 1950. 


Methods of Renal Study 

ASSOCIATE EDiTOR— A. C. Corcoraii 


The tide of publications on the function and structure of the kidney 
and on their interrelationships continues to swell. The interest 
which exists is shown by the fact that methods which only a few 
years ago were only of recondite and occasional use among the 
dwellers in the fabled ivory towers have become almost routines 
of scientific medicine. A scattering of specific procedures has been 
developed which fit only within the narrow bounds of an experi¬ 
mental protocol. We are not here as concerned with these as with 
methods which seem of wider application than is generally realized 
and a few so new that their promise is yet to be fulfilled. 

Those selected are drawn from diverse disciplines and technolo¬ 
gies The methods described through page 213 are applicable in 
animal and clinical physiology. The study of renal tubular excre¬ 
tion in vitro (p. 228) is at the margin between physiology and bio¬ 
chemistry, while electron microscopy of the kidney (p. 234) is 
morphologic. The technique of endogenous creatinine clearance 
(p 215) if an easily reproducible test of renal function whic i wil 
be of interest to those engaged in the scientific practice of chinca 
medicine. The balance of the section is for those inteiested . 
experimental pathology in small animals, especially in rats. 

The coverage is obviously incomplete and may be consideie 
definitely spotty. Those who know the field will recognize that a 
orderly exposition even of a special topic might veiy we 1 exhairst 

hhm lioh open new directions in the study of the k.dney, 




Many attempts have been made to determine renal function from 
the rate of disappearance of intravenously injected test substances. 
Altliough mathematically attractive, most such estimates suffer 
from failures of rapid equilibration of rapidly decreasing peripheral 
venous plasma concentrations with body fluids on the one hand and 
simultaneously collected urine on the other. However, in appro¬ 
priate circumstances, 3 methods can give useful approximations of 
excretion rates, especially of substances which are excreted rela¬ 
tively slowly. 

Dominguez has developed the formulae which apply generally 
to disappearance of substances from the plasma by renal excretion, 
with and without simultaneous extraurinary loss. Robson’s 
methods (p. 139) are aimed at clinical evaluation of specific renal 
functions. Differences in their symbols are: 

Significance Dominguez Robson 

Excretion rate y', y uv 

Plasma concentration x', x P 

Clearance A C 

Further, the symbols V, Vi, and V 2 in Dominguez’ formulation 
represent constants: V in Robson’s concept is an empiric function 
of time (t), explicitly V{t) = a -f- ht, written Vi and V 2 to indicate 
the value of V at times h and 

I. Kinetics of Renal Excretion of Injected Substances 

RAFAEL DOMINGUEZ, St. Luke's Hospital, Cleveland 

The rate of excretion of several substances is a linear function of 

the plasma concentration of these substances. The general equation 
can be written as 

y = V + Ax' 

( 1 ) 

where ?/' is the rate of excretion in mg/min, a:' is the plasma con¬ 
centration 111 mg/100 cc, and p and A are constants. This equation 
can always be reduced to equation 

y Ax (2) 

a form specially desirable when the values y' and x' include a blank 




or an endogenous value for the substance in question (1). If th(‘ 
substance is injected rapidly into the veins and is excreted com¬ 
pletely by renal excretion, then, after diffusion between the blood 
and other body fluids has attained a steady state, the time curve 
of the plasma concentration is a simple exponential of the time 

X = ae~“< 


where o = and V is the hypothetical volume of body fluid in 

which the substance is distributed in 1 fluid compartment at the 
same concentration as in plasma (2). The amount excreted in the 
time ti is 

^ y (It = A J' X (It = A 


The assumption of complete excretion by the kidney makes the 
quantity excreted approximately ecpial to the amount injected G. 
If, in addition, the protluct at\ is so large that the term e can 
be neglected, eciuation (4) becomes 

Therefore, when both utilization and diffusion are neglected, the 
1st approximation (3) to the clearance of the substance by plasma 
data alone is given by 

If diffusion into a 2d fluid compartment is taken into considera¬ 
tion (4), 2 exponentials may be sufficient to express the time change 
of the plasma concentration x 

X = ae~°‘^ + 

and also the time change of the concentration 2 in the 2d fluid com 


Z = c — e-^0 

( 8 ) 

with c given Iiy 


fia +jxh 
/3 — « 


From equation (7) 


2d apiiroximation A 2 to the clearance is 



( 10 ) 

.I2 = 


If the substance is in part excreted by the kidney and in part 
metabolized (utilized), and the substance disappears exponentially 
from the blood stream, another equation is required 

u = Bx 

( 11 ) 

symbolizing the relationship of the rate of utilization u to the 
plasma concentration x, with B a constant (2). 

In this case, the amount excreted E will always be less than the 
amount injected G, and the clearance A 3 is 

leading to 

— (1 — e-a'i) 


( 12 ) 


if the period of collection h is long enough (3). 

In equations (12) and (13) utilization is taken care of, but not 
diffusion. If both utilization and diffusion are taken into considera¬ 
tion (3), 

- (1 - e-«<i) + I (1 _ e-ftt,) (^4) 

« ft ' 

an equation which simplifies to 


a b 

a ft 


if h is long enough. 

The ei ror made in Ai 
the ratio 

by neglecting diffusion is determined by 

— = 1 I “5 
A , ~ 

( 10 ) 

The error made in A, and 
mined by the ratios 

in A 2 by a.ssuming no utilization is deter- 




Finally, the error introduced in Ai by neglecting both diffusion 
and utilization is 

A 4 




1 + 




The volumes of distribution of the substance associated with Ai 
and A 2 (5) can be computed from the ecpiations 


G ^ A 
a tt 


, ^ G{0^a + a V>) 
{I3a + ahA 

( 20 ) 

The error made in Ui, by neglecting diffusion before the steady 
state, can be computed from the ratio (3) 

1 + 2 




( 21 ) 

Note .—Many of the equations given here have been tabulated 
without development in reference (3). The derivation of the equa¬ 
tions for substances which penetrate 2 fluid compartments are to 
be found in reference (4). A review of this work with extension to 
intestinal absorption and a survey of volumes of distribution is pre¬ 
sented in reference (5). 

Comment by James S. Robson 

V (of Dominguez) is taken as the “volume of body fluid in which the 
substance is distributed at the same concentration as in the p^lasma and 
is assumed to be constant in the derivation of formu a (6). The assump¬ 
tion is justified when equilibration of injected material between plasma and 
extracellular fluid is attained at the instant of injection and thereafter 
maintained. If, however, equilibration occurs at some 
and a steady state is subsequently maintained, the assumption is Justifie 
only from that time, and formula (14), incorporating 
before equilibration and for utilization or destruction in the body become. 

valid, while formula (6) ceases to apply. HinHnnP or PAH in 

However following single intravenous injections of diodone or PAH m 
anlounrsitabrJ estimations of renal blood flow, and of inuhn m 



Iarf>;er singlp injpctioiis of iiiulin than used hitherto, equilibration might be 
attained l^efore the substance disappeared from the plasma, although this 
remains to he tested. 

Because of tliese facts, the formulae of Robson et al. were devised. In 
these, the volume of di.strihution is used empirically as a function which 
varies with time, and the experimental conditions in the cases of inulin, 
diodone and PAH are such that the function increases at an approximately 
constant rate between defined limits of time following injection. 


1. Doiuiiigucz, R., and Pomorcne, E.: Studies of renal excretion of creatinine: 

I. On functional relation between rate of output and concentration in 
plasma, J. Biol. Chem. 104:449, li)34. 

2. Dominguez, R.: Studies of renal e.xcretion of creatinine: II. Volume of dist ri- 

bution, Proc. Soc. Exper. Biol. A Med. 31: 1146 and 1150, 1934. 

3. Dominguez, R.; Corcoran, A. C., and Page, I. II.: Mannitol: Kinetics of dis¬ 

tribution, excretion and utilization in human beings, J. Lab, & Clin. Med. 
32: 1192, 1947. 

4. Dominguez, R.; Goldblatt, H., and Pomerene, E.: Kinetics of excretion and 

utilization of xylose. Am. J. Physiol. 119: 429, 1937. 

5. Dominguez, R.: Kinetics of Elimination, Absorption and Volume of Distri¬ 

bution in the Organism, in Clas.ser, O. (ed.): Medical Physics (Chicago: 
Year Book Publishers, Inc., 1950), \'ol. II. 

II. Single Injection Technique in Evaluation of 

Renal Function 

JAMES S. ROBSON, University cf Edinburgh 

Single injection methods of evaluating discrete renal functions 
in human subjects have been suggested by numerous workers, some 
of whom have attempted to simplify the technique of Smith, 
Goldiing and Chasis (13). Simple slope analysis which takes into 
account solely the rate of disappearance from plasma of injected 
material (1, 2, 5) has been shown to be invalid for most substances. 
Such analysis assumes, among other things, instantaneous and 
maintained equilibration of the injected substance in its volume of 
distribution. In the case of mannitol, for examnle. this P.niiiHBrn_ 

inhision c;ui be avoideil. 




lM)ll<>wiTi^ ;i single in<ravenous injection of inulin, diodono and 
PAH in aniouiits suitable lor tlie deterniinatioii of their respectiv'e 
clearances, their plasma concentrations fall with time from the 
instant ot injection. IMeanwhile, their volumes of distribution as 
defined by Robson et al. (9-11) continue to rise. This is because 
plasma concentration does not represent the average extravascidar 
concentration; indeed, the latter changes at a rate which is slower 
than the rate of change of plasma concentration (9-11). In these 

dX dFV 

dt dr 

( 1 ) 

where X = the amount of substance in the body; P = plasma wa¬ 
ter concentration of the substance; V = volume of distribution and 

is and m = rate of loss of substance in the urine. 

It may be shown that V increases approximately linearly with 
time for a considerable period after the injection of inulin, diodone 
or PAH. Equation (1) maybe written 

— nv = (a + + Fh 

where V = a ht, in which a and b are constants. Theretore 

(a + bt) dF 

-C = h + 



( 2 ), 

where C = ^ and i-epresents plasma water clearance and is as¬ 
sumed to be constant. Solving fornuda (2) and integrating between 
2 times, b and r, 

r - KloSio Fi - loKio Fo) _ 
logio Vz — logio Vi 


a) Preparation and injection of solution —Imihn (Kertoot’s pure 
bacteriological) is dissolved in sterile saline with heat to make an 
approximately 10 per cent (w/v) solution, and the solution is 
passed through a sterile Seitz filter. It is then suitable for injection. 
The inulin content of each batch of solution is determined chemi- 

""""A^sample of venous blood (for blank analysis) is withdrawn from 
the subject and an accurately measured volume of the mulin so u- 



tioii, coiiLuninK 100-120 mg of inulin/kg of body m'iglit, is in- 
jRotod intnivoiiously (h syriiigR ol 100 ml (*ap<icity is cmivoniRiit), 
the iiijeclioii taking 2-5 min. The time is noted when the injeetion 
begins and ends. With this technicino, no evidenee of toxicity has 

been observ'cd. 

b) Blood and urine collection. —At 2 noted times (h and 4) about 
30 and 100 min after the midpoint of the injection, urine is col¬ 
lected, the bladder being emptied as completely as possible. Since 
it is necessary to know only the total amount of inulin excreted, 
there is no need for the subject to empty the bladder before the 
injection; on the contrary, a full bladder at that time increases the 
accuracy of the first urine collection 30 min later. Two to 3 min be¬ 
fore these urine collections, venous blood is withdrawn (10 ml). 

c) Calculation. —The weight of inulin injected minus the Aveight of 
inulin in the urine passed at the first collection (h) gives the weight 
of inulin remaining in the body at that time. This, divided by the 
plasma water-inulin concentration Pj, taken 2-3 min before this 
collection, giA’es the first volume of distribution Vy. Similarly, the 
weight of inulin injected minus the total weight excreted in both 
urine samples gives the weight remaining in the body at the time 
of the 2d urine collection {h), and this, divided by the second 
plasma water concentration of inulin Po, gives the second volume 
of distribution Fa. 

C is readily converted to the conventional plasma clearance of 
inulin (Cin) in ml of plasma/min by multiplying by the factor, 
100/(100 — g protein/lOO ml plasma). 


The method is essentially similar to that for inulin clearance and 

so also is the calculation. C X 100 yields estimated plasma clear¬ 

a) Preparation and ejection of solution —It is convenient to 
use, without dilution, the 35 per cent (w/v) sterile solution of dio- 
done obtainable commercially in sterile ampules. The dose required 
IS one which will give plasma-diodone levels of between 5 and 1 mg 
of diodoiie-iodine 100 ml of plasma for the period of the urine col¬ 
lection. I his reciuires 7-8 ml of solution for a person of 70 ko- bodv 
weight and normal renal function; when severe depression of renal 
unction IS expected, the dose should be reduced to 4 ml for 70 ka- 
body «-eight. The injection should take 1-2 min, the exact times 
being recorded. 1 oxic reactions to these small amounts of diodone 

TOmlt"nK.™""’"’ ' *‘'™™us. flushing and 

Solnlions of PAIT may be purchased or prepared by di.ssolving 



PATI acid in normal NaOH, carbonate-free, with subsequent ad¬ 
justment of the pH to 7.2. Three grams may easily bo dissolved in 
15 ml. The solution is then passed through a Seitz filter and auto¬ 
claved. The dose required is one which will give {plasma diodone- 
iodine concentrations between 5 and 10 mg/100 ml for the duration 
of the test. This requires 9-12 ml of the above solution for normal 
people of 70 kg. The amount should be reduced if depression in 
renal function is suspected. The injection should take 2-5 min. 
Toxic effects, including vomiting occasionally occur. 

b) Blood and urine collection .—The procedure is similar to that 
after inulin injection, but because of the rapid elimination of dio- 
done and PAH the first blood and urine collections should be made 
20 min, and the second 60 min, after the midpoint of the injection. 

c) Calculation .—For the respective collection times h and ti 
(measured in minutes from the midpoint of the injection), the vol¬ 
umes of distribution Vi and V 2 are calculated exactly as lor inulin. 
Pi and P 2 are the concentrations of diodone-iodine or PAH in 
mg/100 ml of plasma water at times and t 2 . The constants of 
equation (3) and the clearances in terms of water or plasma are 
also calculated as for inulin. 



A method of analysis similar to that used for inulin aiid diodone 
clearance may be applied to estimate the maximal tubidar excre¬ 
tory power for diodone (10). Following a single intravenous injec¬ 
tion of diodone large enough to saturate tubular excretory capacity 
for the period of urine collection, the plasma concentration falls 
and the volume of distribution rises from the instarit ol injection. 
Accepting maximal excretory capacity as a constant, the basic 
formula of Smith et al. (13), rearianged, is 

uv = Tmn -f FWCinP 

where FWCmP = the filtered diodone (14). Employing symbols 
similar to those used in the preceding section 

_ = (Tmn + FWCinF) 

Now, X 

P V. Therefore 



+ F 






(I'VVCi- + jf) " 

- U 
‘ dl 


Under the conditions of the test it may be sliown that V varies 
lineai’ly with time. Therefore 

Solving this for P ynd integrating between times ii and I 2 , 

C' + b C' + b 


Vi ^ - V2 ^ 

where C' == FW^C,„. 


а) Injection of diodonc .—The required amount of diodone ranges 
from 350 to 200 mg/kg of body weight according to t lie (expected) 
i-enal efficiency of the subject: the greater the efficiency tlie greater 
(he dose. The solution used, 50 per cent diodone (w/v), is obtain¬ 
able, sterile, in ampules, and an average-sized normal subject is 
given 30-50 ml at a rate not exceeding 5 ml/min. 

When given in these quantities, diodone occasionally produces 
side-reactions which include nausea, vomiting, tenesmus, colic, 
urticaria, faintness, fall in blood pressure, itching and pain over the 
injection site. Incidence and severity of such reactions are, how¬ 
ever, minimized by the prescribed slow rate of injection. 

б) Collection and oncdysis of samples and calculation. —Measur¬ 

ing time, as usual, from the midpoint of the injection, blood and 
urine are collected after approximately 30 and 100 min, Vi and V 2 
aie calculated, using Pi and P 2 as described, and thence, as before 
the constant 6. ’ 

Pi and P 2 are the plasma water concentrations of diodonc-ir»lit,o 

>t determining discrete renal func- 
ivenous infusion and collection of 



urine by catheter over short periods (10-15 miii) are freer from 
tlieoretical objections and the possibility of systematic errors than 
single injection techniques. Nevertheless, any method which ex¬ 
tends the possible period of urine collection to 60-80 min succeeds 
in eliminating the greatest single source of technical error in such 
determinations, namely, the direct estimation of excretory rates 
by sampling of voided urine over short periods. 

In practice, the differences between results of the methods de¬ 
scribed and those involving continuous infusion with short period 
urine collection are insignificant as far as inulin clearance and 
maximal tubular excretory capacity are concerned (6). This is be¬ 
cause arteriovenous differences in concentration of inulin and dio- 
done with single intravenous injections of appropriate magnitude 
are insignificant for inulin (4) and of the order of 5-10 per cent for 
diodone (11). Agreement between estimated diodone and PAH 
clearances as determined by the 2 methods is not as good; the 

infusion method gives results which average 20 per cent higher 
than those obtained by single injection. The arteriovenous differ¬ 
ence of diodone concentrations in these circumstances varies from 
7 to 25 per cent and is apparently roughly proportional to the level 
of renal blood flow (6). Inequality between peripheral venous and 
mixed arterial concentrations is largely responsible for the dispari¬ 
ties in clearance values obtained by the 2 methods. Ihis disadvan¬ 
tage of the single injection method can be overcome either by sam¬ 
pling arterial blood or by setting up an arbitrary range based on 
venous sampling at various levels of renal blood flow. Development 
of micromethods applicable to capillary blood might effective y 
replace the former. Still, venous sampling permits analysis of data 
on values thus derived which correspond with similar data obkiinetl 
by infusion techniques at least in all types of abnormal renal con¬ 
ditions which have been studied (7, 11). 

There have been other recent objections to the single injection 

technique (12). Variations in the volume “t 
over the period in which the formulae are applied do not « 

validity of the method, in the sense that they yield values ot cle<. - 
aiice w'hich are the averase for the time over winch ca ™- 

lated Abnormalities in the value for the volume of distribution 

|:rent)'wS' -IV 

rate of mixing between these compai tments. 



There does remain, however, the possibility that error might 
arise from the difficulty in relating any particular amount of ex¬ 
creted substance to a plasma concentration which is changing witli 
time (12). This does not appear to be significant in the case of 
inulin clearance and maximal tubular excretory powder (6). Its 
effect is likely to be minimized, moreov^er, when long periods of 
urine collection are related to a large segment of the plasma con¬ 
centration-time cur\^e, for the reason that increments or deficits in 
the excreted substance which might result from such a cause will 
be smaller in relation to the total collected than Avith short collec¬ 
tion periods. Nevertheless, it seems likely that a real and significant 
error is introduced into all single injection methods in which rapid 
change in plasma level occurs, i.e., in determinations of diodone 
and PAH clearance, as a result of this failure to sample representa¬ 
tive urine. The magnitude of this error is unknown. It is, however, 
not likely to be large since results of single injection clearance 
methods closely approximate those obtained by infusion methods 
when arterial blood is sampled and when the customary 2-3 min 
allowance is made for delay or minimal appearance time. 

Note.— This section was leviewed by Rafael Dominguez. 



1. Alving, A. S., and Miller, B. F.: A practical method for the measurement of 

glomerular filtration rate (inulin clearance) with an evaluation of the clini¬ 
cal significance, Arch. Int. Med. 66: 306, 1940. 

2. Barnett, H. L.: Renal physiology in infants and children: Method of esti- 

*^4^^T940 filtration rate, Proc. Soc. Exper. Biol. & Med. 44: 

Dominguez, R.; Corcoran, A. C., and Page, f. H.: Mannitol: Kinetics of 

Med! 32:Tl9^794^^ J- & Clin. 

4. Ferguson, M. H.’; Olbrich, O.; Robson, J. S., and Stewart, C. P.: The use of 
25t"l95o‘" filtiation. Quart. J. E.xper. Physiol. 35: 

6. Olbrich, O.; Ferguson, M. IE; Robson, J. S., and Stewart C P • A « 
parison of the continuous infusion and single injection methods* 

7 renal functions, Edinburgh M. J, o7' 110 950 ' 

u); I'iso.' ■" l-^n,"“uVgh M. ,1. 57: 

of inulin a’,,,I diodm!"! lOTt '<'"••1 cleuranco 

o lo K.> -'OO, i\ov. 18 , lu.'AO 


10. Hobson, J. S., el al.: The determination of the renal clearance of diodone 

and the maximal tubular excretory capacity for diodone in man, Quart. J. 
Kxper. Physiol. 35:173, 1949. 

11. Robson, J. S.; Horn, D. B.; Spreull, E., and Stewart, C. P.: Unpubli.shed 

12. Smith, n. W.: The Kidney: Structure and Function in Health and Disease 

(New York: Oxford University Press, 1951). 

13. Smith, H. W.; Goldring, W., and Chasis, H.: The measurement of tubular 

excretory mass, effective blood flow and filtration rate in the normal hu¬ 
man kidney, J. Clin. Invest. 17:263,1938. 

14. Smith, W. W., and Smith, II. W.: Protein binding of phenol red, diodrast 

and other substances in plasma, J. Biol. Chem. 124:107, 1938. 


I. Extraction and Clearance Method 

JAMES F. NICKEL and STANLEY E. BRADLEY, Columbia University 

The use of sodium p-aminohippiiriite or Diodrast clearance for 
the measurement of renal plasma flow is based on the assumption 
that these substances are completely cleared from the plasma 
perfusing the kidneys. A\ hen renal extraction of PAH or Diodrast 
is impaired by tubular dysfunction or by competitive interference 
by other substances excreted in a similar fashion, the clearance 
procedure alone is insufficient for this purpose and the actual 
amount of PAH or Diodrast removed from each volume of plasma 
flowing through the kidneys must be measured directly. Develop¬ 
ment of the venous catheterization technique by Cournand and his 
associates (2) has made this possible in man and animals under 
physiologic conditions. 


Under local anesthesia,* through a small incision a radiopaque 
ureteral catheter is introduced into a medial antecubital vein and 
passed under fluoroscopic control by way of the superior vena 
cava, right atrium and inferior vena cava into the right renal vein. 
(1 he right renal vein must be used because the ovarian or sper¬ 
matic veins empty into the left renal vein and peripheral venous 
blood is mixed with renal venous blood.) Isotonic saline or glucose 
solution is allowed to flow slowly through the catheter while it is 
kept in place. To avoid dilution by the infusion fluid 2 or 3 ml 
of blood is withdrawn and discarded before a sample is taken for 
analysis. As a rule, blood samples to serve as blanks are taken simul¬ 
taneously from a peripheral vein and the renal vein before deter¬ 
mination of clearances. Clearances are measured as described by 
Selkurt (3) and reiial venous blood is sampled simultaneously 
with peripheral venous blood. 

The total renal plasma flow may then be calculated by dividing 
the amount of the material excreted in the urine each minute, 
L F, by the difference between the concentrations in each milliliter 
of peripheral and renal venous plasma or by dividing the clearance 

* Procaine should not he used if PAH extraction 
lere with the chemical deterniination of P.\n. 

is measured 

because it may inter- 




value C, by the renal extraction ratio E (the concentration differ¬ 
ence between peripheral and renal venous blood divided by the 
peripheral concentration). It can be seen that any substance which 
is extracted from the blood by the kidney and excreted as such may 
be used for this purpose. Inulin, mannitol and urea as well as 
PAH and Diodrast have been employed at various times. Diodrast 
is seldom, if ever, used because it enters the erythrocytes and dif¬ 
fuses out into the plasma once it has been removed from the plasma 
by the renal tubule cells. Rapid separation of plasma and red blood 
(*ells may circumvent this difficulty, but it is preferable and more 
convenient to use PAH in man because it is not taken up by the 
erythrocytes and is easily determined chemically. There is no 
evidence that PAH is appreciably acetylated or metabolized by the 
human kidney (1). 

Comment by J. K. Clark 

The technical details of renal venous catheterization are much the same 
as those of other catheterization methods (4). The patient should be su¬ 
pine. Since catheters sterilized by boiling soften, they may be stiffened in 
iced saline before use. Venous spasm which may interfere with effective 
catheter manipulation can often be relieved by such means as stopping 
movement of the catheter, warming the arm, infiltrating a local anesthetic 
(choosing one which will not interfere with chemical determinations) along 
the vein, withdrawing the catheter and reinserting it after an interval or 

inserting a smaller catheter. 

On insertion the catheter may tend to pass into the neck, the light ven¬ 
tricle or a hepatic vein, so that passage into the right renal vein is a matter 
of trial and error based on experience and confidence. Entry into the vein 
is usually obvious fluoroscopically. Respiration and softening of the cathe¬ 
ter may result in displacement, so that the position should be checked as 
the experiment ends. Difficulty in withdrawing blood may result from en¬ 
try into one of the small branches of the renal vein or pressure of the cathe- 
ter against the vein wall. Careful retraction under observation will correct 
this. Deep inspiration sometimes facilitates sampling but may lead to is 
placement and alter the results of gas analysis. , font 

Catheters through which many samples are to be taken 
bv adding a small amount of heparin to the saline drip feince lenal venous 
Wank Xs for inulin, mannitol and PAH are -‘I* pe;';phem 

arterial or venous blood, time can be saved by taking peripheial blood fo 
bilnk a™d starting toe infusion of test solution he^e cat — 
begins. The desired stable plasma concentrations are theieb> attained 

"Tte“STntutled^^^^ withdrawal, A pressure dressing will 



diuical contraindication to this procedure. However, we have avoided its 
use in subjects with myocardial irritability or a tendency to thrombosis. 

Contrary to earlier impressions, recent evidence indicates that PAH may 
penetrate red cells in man. In a communication to Dr. H. G. Barker, Dr. 
R. A. Phillips states that although the process is slow, it is nevertheless defi¬ 
nite, and in unusual circumstances, such as elevation of plasma PAH 
concentration for measurement of Tm followed by a fall in concentration, 
diffusion from red cells may invalidate the use of plasma concentrations in 
the Pick equation. Conn and Markley have matched Pick PAH determina¬ 
tions against direct (bubble flow meter) measurements of renal blood flow 
and have demonstrated the validity of the equation as applied to whole 
blood. Unfortunately, determination of PAH in whole blood is less satis¬ 
factory than in plasma. Under usual conditions, especially if plasma is 
separated with reasonable promptness, red cell diffusion of PAH should 
introduce little error into estimates of blood flow based on plasma deter¬ 
minations, since a false doubling of the renal venous estimate will decrease 
apparent blood flow by less than 10 per cent at normal levels of extraction 
ratio. This factor becomes serious as extraction falls. 

Comment by A. C. Corcoran 

Wolf (5) pointed out that renal extraction of water should be allowed for 
in calculations of renal blood flow based on renal arteriovenous difference. 
This allowance increases at high levels of urine flow, as during osmotic diu¬ 
resis, and when the extraction ratio measured is small. However, in general, 
such errors are small as compared with technical inaccuracies. Thus, in the 
case of inulin, 5 per cent errors in opposite senses in arterial and venous 
plasma concentration estimates will alter the estimate of extraction from 
0.2 (true) to 0.28 or 0.105 (apparent); at an inulin clearance of 130 cc/min, 
the true renal plasma flow of 6.50 will be recorded respectively as 465 or 
1,240 cc/min. The importance of scrupulous technique and of Using esti- 
m^es of blood flow on substances of high extraction is therefore apparent. 

The technique of determination of renal blood flow by measurement of 
clearance and extraction is described not only because of its intrinsic im¬ 
portance in the study of renal blood flow in the diseases of human beings 
but because the process of catheterization of the renal vein will probably 
come into still wider use in the study of renal metabolism. 


^ function in renal diseases, Am. J. Med. 9: 766, 

2 . Cournand, A F., and Ranges, H. A.: Catheterization of the right auricle in 
man, Proo. Soc. Exper. Biol. & Med. 46: 462, 1941. 

S. Selkiirt E. E.: Measurement of Renal Blood Flow in Potter V T? v 

IXXrizX,fires' V R M 

.;ugo: Year Hook Pul,lii",Wluo/,t8V VoM 

Am:j »">■ "“'v or extrar-ti,,,, ration, 




II. Methods for Direct Measurement 

EWALD E. SELKURT, Western Reserve University 

One of the desirable attributes of a directly recording flow meter 
would be that it offer minimal resistance to blood flow so that 
nearly normal pressure relationships would be obtained whether 
used in series with the artery (arterial inflow method) or vein 
(venous outflow method). Although the optically recording bubble 
flow meter is unsurpassed for simplicity and precision (4), it suffers 
from the disadvantage of not being continuously recording and, 
in the range of renal blood flows, causes considerable pressure 
drop because of its resistance, hence making it unsuitable for 
venous outflow measurement. To obtain continuous recording of 
renal blood flow with minimal pressure loss, use has been made 
of a large size optically recording rotameter originally designed for 
direct measurement of cardiac output in dogs (2, 6). 


This rotameter has a diameter of 30 mm across its chamber and 
contains a float of 22 mm diameter. Different ranges of flow are 
obtained by using floats of different weights: Lucite, 0-500 cc/min, 
Lucite with brass slug, 0-900 cc/min; all brass float, 0-1,500 cc/min. 
The 1st range finds best application for renal work. Pressure drop 
across the rotameter and connections has been found to be about 

4 mm Hg at 100 cc/min, and about 9 mm Hg at 200 cc/min. The 
low resistance of this meter makes it permissible to use it as a 
venous outflow recorder as well as for arterial input meteiing. 
Cannulation of the vein can be direct, but it has been used m 
connection with indirect cannulation of the left vein via the right 
iugular vein and inferior vena cava. A suggested modification ot 
the original method (3) is to use plastic tubing of 4 mm ID and 

5 mm OD guided into the vein by a loose-fitting stilet (1). Blood 
nassing out through the rotameter is passed directly into the oppo- 
SuLlar vein. By projection of the return cannula into the supe- 
rifr vena cava, advL^tage can be taken ot the “ n, 
pressure to compensate to a large degree for pressuie diop acioss 

the circuit. 


Connections tor using the rotameter for arterial mflow measure¬ 
ment as used in this laboratory are shown in Figiire 1. the Wt 


through an external circuit of Tygon tubing from the carotid 
artery. The rotameter is in series with this circuit. The heparinoid 
Paritol* has been used successfully as an anticoagulant in initial 
dosage of 25-50 mg/kg intravenously, with booster doses of 50 mg 
total every half-hour. A pumpt is used to obtain a wide range of 
perfusion pressures and, in connection with 2 damping chambers, 
Di and Di, also provides a wide range of pulse pressures. The 

Fig. 1. —Perfusion circuit for measurement of direct renal blood flow using 
rotameter principle with provisions for varying mean perfusion pressure and 
pulse pressure. Rotameter calibration is achieved by shunt circuit to the fem¬ 
oral vein and an electrically recording 3-way stopcock, S. During calibration 
the kidney is supplied from the femoral artery {lower right) by opening clamp 9 
and closing clamp 8. Calibration flows are varied with clamp 7. Pump chamber 
A, IS ^^de of Elastomold tubing of 3.5 cm OD and 12 cm length and supplied 
with rubber flap valves at top and bottom. It is actuated bv a rotating ball¬ 
bearing cam driven by a Bodine motor turning at 120 rpm. Motor is mounted 
on a worm gear, B, so that stroke volume can be altered. Dj and are air 
damping chambers. Clamps 1 and 4 are used to close off the pump circuit- 

clamps ^ and 6 aid in adjustment of mean and pulse pressures: clamps S and 5 
regulate the air chambers. o. a, .u o. 

advantage of this circuit is that the rotameter can be calibrated at 
any time by supplying the kidney by an alternative circuit from 
the femoral artery by opening clamp 9 and closing clamp 8 . In the 
meantime blood is passed through the rotameter and shunted 
back through the femoral vein at varying rates dependent on the 
setting of clamp 7. For calibration after flow has stabilized, about 
10 cc IS shu nted into a cylinder from a 3-way stopcock, S, with elec- 

Phufdelphia. Institute of Applied Biochemistry, 

ji^t The pump was constructed by Mr. Clifford Wilson. 325 W. 42d St.. Indianapo- 



trical contacts for precise recording of the time of outflow by means 
of a signal magnet registering on the optical record. 

Critique .—Difficulties encountered in use of this rotameter 
have been the result of the nonlinear character of the calibration 
curve, necessitating numerous determinations for best accuracy, 
and the tendency of the curve to drift. The former is due to inherent 
characteristics of the electrical circuit, and the drift is due to tem¬ 
perature change. Changes in blood viscosity, possible fibrin for¬ 
mation and changes in line voltage may be other sources of diffi- 

Pjq 2.—4 -S model recording rotameter, 400 ml capacity. (Courtesy of Shipley 

and Wilson (5).) 

CLilty. As pointed out by Shipley, the instrument is a mean flow 
recorder and is capable of quantitating pulsatile flows with rea¬ 
sonable accuracy only when the float does not hit the top or bottom 
at the extremes of its excursion, and only when entering fluid is 
not directed as a jet into the bottom of the instrument. The latter 
occurs when adapters or tubing smaller than the lnflo^^ spout 

used to convey blood to the rotameter. i i cu* i r 

Thriatest fype model, 4-S series (Fig. 2), designed by Shipley 
.and Wilson (5) attempts to minimize some oi these disadvantages 
of the older models. The improved features ol the 

1 A standard detecting top portion which may be used 


interchangeable bottom metering portions having different flow 

2. Inclusio?! in the detecting lop portion of a noniiidiictively 
wound coil (made of the same wire as the detecting coil) which 
serves as the opposing resistance arm of the bridge. This change 
essentially eliminates any drift due to changes in blood tempera¬ 
ture causing change in coil resistance. 

3. Use of manganin or Ad\'ance wire wound resistors inside tlie 
control box for the other 2 arms of the bridge. Additional thermal 
drift is eliminated thereby. 

4. Widening of the Lucite chamber in which the float rides so 
that for all metering bottoms the pressure drop required to just 
lift the float is between 1.5 to 2.0 mm Hg. Pressure drop across the 
entire instrument is estimated to be only 2.0-2.5 mm Hg at maxi¬ 
mal flow. 

5. Redesign of the float disk so that changes in viscosity of the 
blood have a minimal effect on the height of the float. 

6. The calibration curve represents a better approximation to 
linearity in that it is the central or least curvilinear portion of a 
sigmoid curve. 

Comment by R. E. Shipley 

In common with other types of direct flow measuring devices the record¬ 
ing rotameter causes a relatively small but definite hindrance to the flow of 
blood. By increase of the bore of the rotameter and diameter of the float the 
pressure drop across the instrument could be reduced to an almost negli¬ 
gible value; however, the increased volume of blood required to fill the en¬ 
larged rotameter chamber might constitute a disadvantage in some experi¬ 

As a supplement to Dr. Selkurt’s suggestions, it should be stressed that 
the accessory conducting tubing, connections, cannulae, etc., be made as 
short as possible and of maximal bore since the resistance to flow through 
the entire circuit determines the pressure drop and therefore the extent to 
which the flow of blood will be hindered. 

Comment by A. C. Corcoran 

The procedure described here is in a sense supplementary to Dr Sel- 
kurt s contributions to this topic in Volume 1 of this series (pp. 191 ff) 

developed largely since that time; as the 
characteristics of rotameters become more widely understood, the ingeni¬ 
ous instruments are being put to use far afield from the kidney. ^ 


'■ - the inta..t 



3. Selkurt, E. E.: Comparison of renal clearance with direct renal blood flow 

under control conditions and following renal ischemia, Am. J. Physiol. 145: 
376-386, 1946. 

4. Selkurt, E. E.: An optically recording bubble flow meter adapted for meas¬ 

urement of renal blood flow, J. Lab. & Clin. Med. 34:146-150, 1949. 

5. Shipley, R. E., and Wilson, C.: An improved recording rotameter, Proc. 

Soc. Exper. Biol. & Med. 78: 724-728, 1951. 

6 . Study, R. S., and Shipley, R. E.: Comparison of direct with indirect renal 

blood flow, extraction of inulin and diodrast, before and during acute renal 
nerve stimulation. Am. J. Physiol. 163:442-453,1950. 




The techniques described here are so diverse that they fall 
under one heading only by a rather generous interpretation of the 
function of classification. Deane’s description of infusion technique 
should perhaps have followed Dominguez’ and Robson’s contribu¬ 
tions on the estimation of function from plasma disappearance 
rates had it not run so naturally into and along with his paper on 
body water compartments (space techniques). The latter supple¬ 
ments parts of the section on Fluid and Electrolyte Distribution 
edited by Flexner in V^olume 4 of this series; it also carries on very 
naturally into the discussions by Wesson and Brodsky of the rela¬ 
tionships between kidney function and the fluid, electrolyte and 
nonelectrolyte masses of the body. Experiments can readily be 
visualized which would incorporate some of Deane’s methods with 
some of Wesson’s studies in the dog. Brodsky’s contribution is in 
part complementary to Wesson’s discussion of osmotic diuresis 
and in part a new approach to the study of renal function in the 
diseases of human beings. 

The transition from osmotic diuresis to the assay of diuretic and 
antidiuretic substances is perhaps easier etymologically than 
physiologically. The contributions by Little, Turner and Grollman, 
and Gaunt form substantially as complete a survey of this still 
vexed field as could be gathered. 

Renal e.xcretion of water and electrolytes can be assessed either in 
teims of over-all body economy as, for example, in the studies 
elaborated by A. V. Wolf in his excellent text, The Urinary Func¬ 
tion of the Kidney (2), or as discrete and isolatable functions of 
nephrons and parts of nephrons. Both approaches have excellencies 

and advantages, the latter illustrated by Homer Smith in The 
Kidney (1). 

Among discrete functions, the excretions of water and salt are of 
e first interest, and perhaps the greatest complexity, and their 
raploration is only beginning. For these reasons, Wes.son and 

fom?d’'t''-rV'"''‘®n ‘‘T'"'’ P'-o^dures and concepts in 

■ ome detail. In so doing, they have brought to light aspects of 

conceptual disagreement which future studies will Resolve These 




arise in part from inherent differences in their aims. Wesson’s arc 
more academically and Brodsky’s more clinically oriented. Wesson’s 
hope, and ours, is not as much to encourage exact repetition of his 
carefully controlled experiments as to enable the thoughtful reader 
to proceed with modifications and elaborations. Brodsky and his 
colleagues have specified a measurable aspect of kidney function 
which can be studied in dog and man, while Wesson points out 
that the rigors of urea diuresis can hardly find application in human 
beings. Both agree that the application of the methods they de¬ 
scribe to studies of renal responses to drugs and to loads, in health 
and disease, have yet to be developed. The methods are presented 
not as definitive procedures with settled applications but as illus¬ 
trations of a new and important direction in renal physiology. 


1 . Smith, H. W.; The Kidney: Structure and Function in Health and Disease, 

(New York: Oxford University Press, 1951). 

2. Wolf, A. V.: The Urinary Function of the Kidney (New York: Gmne & 

Stratton, Inc., 1951). 

I. Infusion Technique for Measurement of 
Renal Function 

NORMAN DEANE, Neio York University 

Earle and Berliner (3) have described an infusion technique for 
measurement of renal function without bladder catheterization or 
collection of urine. Comparisons of the infusion and urine col ec- 
tion methods have been presented by Berger, Farber and Earle (1), 
who carefully defined the principle on which the proceduie is base . 
“For any exogenous substance which is neither metabolized, nor 
e4rete7othenvise than in urine, the rate of excrct.on must be 
equal to the rate of infusion under conditions 
and volume of distribution are constant. This steadj state 

time onnS'“^ the rate at th-uh- 

Stance is being excreted, UV (mg/min), is equal to the rate at 
thich it is being injected, IV (mg/min), and the clearance calcula- 

tiori can be stated 


1 c .Y. T = concentration of x in 


concentration of x (mg/cc). Inulin and p-antiinohippuric acid 
clearances may be determined by this technique.* 


1. A blood sample is obtained for determining plasma blank, and 
priming injections of inulin and PAH are administered. The prim¬ 
ing injections are calculated as follows: 

Priming in- = Pwgt of sub-1 fvol of distribu-1 Tintended pla.smal f j qqqI 
jection (mg) L ject (kg) J L tion (% wgt) J L concn. (mg/cc) J L ’ J 

For inulin, a volume of distribution of 18 per cent and a plasma 
concentration of 20 mg per cent may be used. For PAH, a volume of 
distribution of 28 per cent is employed and an intended plasma 
concentration of 2 mg per cent is used for measurements of effective 
renal plasma flow. 

2. A sustaining infusion of inulin and PAH is begun (preferably 
through intravenous plastic tubing (4)). For measurements of 
glomerular filtration rate and effective renal plasma flow, the cal¬ 
culation is 

I plasma concn. 1 f estimated clearance! 

Concn. in sustaining _ L (mg/cc) JL _ (cc/min) _J 

infu.sion (mg/cc) ~ [rate of infusion (cc/min)] 

It is extremely important that the infusion rate be kept very 
constant and be known accurately. 

3. After the infusion has continued for 4 hr, a minimum of 3 
blood samples is collected at 30 min intervals, and the infusion is 

Analytical technique. The determination of inulin by the resor¬ 
cinol method is described on page 163. Determination of inulin by 
the diphenylamine reagent and the analytical procedure for PAH 
have been presented by Selkurt (5). 

In the infusion method for measurement of renal function, it is 
necessary to determine ‘Total” plasma PAH after hydrolysis 
since PAH is conjugated in man. This is done by adding 1 cc of 
I.IN HCl to 10 cc of plasma filtrate and heating for 1 hr in a boiling 
water bath One cc of N HCl is added to 5 cc of the hydrolyzed fil¬ 
trate and the color reaction then performed in the usual manner 
Lalcidation.~T\xQ clearance is determined, at the time of infu¬ 
sion equilibrium, as the rate of infusion {IV) divided by the plasma 
concentration. ^ 

If infusion equilibrium as defined by Berger Farber 
and_E^e (1) is reached, the infusion method is valid for measuring 

amoun/of pS Xired"makes^-Thf‘'^ technique. However, the large 

collection method i.s recon,men(ied^fLTeTsu'remen£'ofT^^^^ therefore the urine 


renal function. Inaccurate data are obtained if meticulous attention 
is not directed to the following details and limitations of this 

The infused substance must have reached a maximal volume of 
distribution before calculation of clearance is made, and the plasma 
concentration must be constant. Thus, it is not practical to attempt 
application of this method in instances of increased extracellular 
fluid volume. 

The composition of the infusion and rate of infusion must be 
known exactly. An independent check on the rate of the pump, 
similar to that presented for the calibrated infusion method for 
measurement of extracellular fluid (2), should be maintained 
throughout the period of infusion. An inconstant pump rate 
invalidates the method. 

The use of substances which are not biologically inert (i.e., 
mannitol) invalidates the method. 

The method should not be used for quantitative studies relating 
to acute changes of renal function. Calculations made by the infu¬ 
sion method indicate the direction but not the actual extent of 
sudden alterations in clearance (1). The discrepancy between the 
infusion and the urine collection methods will exist until the change 
in the concentration which wdll be effected by the change in clear¬ 
ance has been communicated throughout the volume of distribu¬ 
tion of the substance. 

Comment by David P. Earle, Jr. 

The importance of strict adherence to the technical details of this 
method cannot be overemphasized. If the concentration of the substance 
whose clearance is being measured varies by more than 5 per cent in the 
plasma samples, infusion equilibrium has not been achieved and the results 
Siould be discarded. Unfortunately, when renal function is less than 60 
per cent of normal, the infusion technique may yield erroneously high 
values even though the plasma levels may indicate an apparent infusion 
equilibrium. This may occur when the sustaining infusion has been 

::if materials in the sustaining 

infusion by the same methods utilized tor the plasma since some prepam- 

“ A« a It CtbriuSf" ?h1 S'om conservative, infusion 

rmeXd bt imiKissible to obtain ac- 

curate urine collection. 



1. Berger, E. Y.; Farber, S. J., and Earle, D. P., Jr.: Comparison of the 

constant infusion and urine collection methods for the measurement o 
renal function, J. Clin. Invest. 27: 710, 1948. 

2. Deane, N.; Schreiner, G. E., and Robertson, J- S.: The velocity of distribu¬ 

tion of sucrose between plasma and interstitial fluid, with reference to 
the use of sucrose for the measurement of extracellular fluid in man, 
J. Clin. Invest. 30: 1463, 1951. 

3. Earle, D. P., and Berliner, R. W.: A simplified clinical procedure for meas¬ 

urement of glomerular filtration rate and renal plasma flow, Proc. Soc. 
Exper. Biol. & Med. 62: 262, 1946. 

4. Ladd, M., and Schreiner, G. E.: Plastic tubing for intravenous alimenta¬ 

tion, J.A.M.A. 145: 642, 1951. 

5. Selkurt, E. E.: in Potter, V. R. (ed.): Methods in Medical Research (Chicago: 

Year Book Publishers, Inc., 1948), Vol. 1, p. 191. 

II. Methods of Study of Body Water Compartments 

(Space Techniques) 

NORMAN DEANE, New York University 

Present methods for study of body water are based on a dilution 
principle. In the case of a substance restricted to a single compart¬ 
ment, the greatest volume of dilution is, of course, equal to the 
volume of the compartment. This volume (the volume of distribu¬ 
tion) is calculated by dividing the amount of the substance present 
in the compartment by the concentration of substance per unit of 
water in any part of the compartment from which a sample may be 

Certain fundamental conditions must be fulfilled in this calcula¬ 
tion. (1) The total amount of the substance present in the compart¬ 
ment must be accurately known, so the method should include 
means of quantitative correction for significant loss through 
excretion or metabolism. (2) The method should have an internal 
check which will indicate that in each measurement complete dis¬ 
tribution throughout the particular compartment has been attained. 
(3) The concentration of the substance should be the same through¬ 
out the compartment at the time of sampling. The nature of the 
substance administered and the physiologic state of the subject will 

determine the procedures to be used if these conditions are to be 


The volumes of distribution of inulin and sucrose more closelv 
approximate the extracellular fluid volume in man than do the 
volumes of distribution of test substances formerly used (5 14) 
They are biologically inert in man and can be quantitative y re- 


covered in the urine after intravenous injection. Because of rapid 
excretion, true plasma-interstitial fluid equilibrium does not occur 
after a single intravenous injection (28), so that continuous infusion 
techniques have been devised to insure complete distribution. 
These substances may be used in either of the methods described 


This method (13, 32) for measurement of extracellular fluid in 
dog and man involves the use of a priming injection of inulin and 
an approximately constant rate of infusion to maintain a constant 
plasma concentration during the interval required for inulin to 
attain uniform distribution throughout the extracellular fluid. 
The duration of the infusion, in the absence of increased extra¬ 
cellular fluid, is 2 hr in the dog and 5 hr in man. At the end of this 
time a plasma sample is obtained, the bladder emptied and the 
infusion discontinued. The urine is collected for 5 hr in the dog and 
12 hr in man. The inulin volume of distribution is equal to the 
amount of inulin recovered in the urine (after correction for urine 
inulinoid blank) divided by the concentration of inulin in plasma 

a) Procedure. —1) The subject is hydrated so that the urine flow 
is 2-A cc/min while the sample for the rate of excretion of inulinoid 
blank is collected and also when the final bladder wash-out is per¬ 
formed. * 

2) The subject is catheterized with a multiple-eyed catheter 
and a timed urine sample is collected lor determination of the rate 
of excretion urine blank {UqV). 

3) A control blood sample is obtained for determination ol plasma 
blank concentration {Bo). 

4) The inulin priming injection is administered. The priming 
injection is calculated assuming an inulin volume of distribution of 
18 per cent of body weight and an intended plasma concentration 

of 20 mg/100 cc. 

Inulin priming injection (mg) = [wgt of subject (kg)][0.18][0.2][l,()00] 

5) The inulin sustaining infusion is started after the priming 
injection is completed. An infusion rate of 0.5-1.0 cymin is pre¬ 
ferred Concentration of inulin in the .sustaiiung infusion is c 

culaled on (he basis of infnsion lale, intended 

tion (20 mg/K)0 cc) and estimated rate of glomerulai filtiation. 

* The technique of bladder wash-out is described by Selkurt in 
Methods in Medical Research, page 191. 



restimated filtration"! Tintended plasma"! 
hmliu concm., ii. »uslain- , L rate (co/mm) JLconcen (mg/ccj j 
ing infusion (mg/cc) [infusion rate (cc/min)] 

6) The infusion is continued for the required length of time and 
lilood samples taken to check the constancy of the plasma concen¬ 
tration. A final blood sample is obtained before the infusion is dis¬ 

7) The bladder is emptied and washed with saline and air. 

8) The infusion is stopped on completion of the bladder wash¬ 

9) All urine is collected for the next 12 hr. 
b) Calculation: 

Inulin vol of distribution (cc) = 

t total amt of inulin"! _ Prate of excretion"! fduration of urine | 
recovered (mg) J Lof UqT (nig/min)J L coll, (min) J 
[inulin concen. in plasma water (mg/cc) 1 


The novel feature in this method of measurement of extracellular 
fluid in man, using sucrose (5), is the calculation of the virtual 
volume of distribution at any instant after the start of an ex¬ 
tremely constant intravenous infusion without use of a priming 

The virtual volume of distribution of any nonmetabolized solute 
X at any time t is given by the expression 

2/F - Sf/F 

where is the virtual volume of distribution (cc), 27F the tota 
quantity of solute that has been infused (mg), St/F the tota 
quantity of solute that has been excreted in the urine (mg), and P 
the concentration of the substance in the plasma (mg/cc). The 
value for F^ increases and approaches a maximal volume which is 
identical with the volume of distribution expected at equilibrium, 
i.e., when P has been maintained constant for some period of time! 
With sucrose, that value of F« which remains constant with pro¬ 
longation of the infusion is taken as the extracellular fluid volume 
An infusion of 3 hr is recommended in the nonedematous adult 
although maximal volume of di.stribiition is usually reached in a 
shorter neriod. 

It IS of critical iiiiportaiice in this method that the coiniiositloii 
ol the iiiiiision he known with great accuracy and that the rate of 
intusion be kept very constant and be known exactly, because slight 
errors in IV compound with time in the calculation of ZIV and 



therefore, in all derivative calculations. We have infused a 9.5-10.5 
per cent sucrose solution with a Bowman constant infusion pump, t 
delivering the infusion fluid at a rate of 0.665-0.700 cc/min. When 
this pump is powered through a Sola constant voltage transformer, 
the rate of infusion, as checked at 30 and 60 min intervals, remains 
constant with an accuracy of ±2 per cent. The infusion reservoir 


] —Typical experiment showing the virtual volume ot distribution of 
sucrose (E») during infusion at a constant rate. 

consists of a series of 20 and 40 cc glass bulbs with calibrated stems 
connected by ground-glass joints, permitting the rate of the in u- 
sion to be checked from time to time. The volumes delivered by the 
bulbs are reproducible within ± 1 per cent despite repeated auto¬ 
claving. Care must be taken to keep them chemically clean so that 
drainage will be complete. It is essential that a continuous check ot 
this type be kept on the rate of the pump throughout the infusion 
a) Procedure.—\) The subject omits breakfast and is hydiated 

"TMai^ufa^tured by Process and Instruments Corporation. Brooklyn. 


to produce a urine flow greater than 3 cc/min throughout the infu¬ 

2) The subject is catheterized with a multiple-eyed catheter. 
The bladder is emptied. 

3) A control blood sample is obtained for determination of 
plasma blank concentration, and plastic intravenous tubing is 
inserted into the vein (18). 

4) The exact time is noted and the infusion started through the 
plastic tubing. The volume infused is obtained by noting the exact 
time at which the fluid level passes the calibration marks on the 
infusion apparatus. 

5) Timed urine samples are collected at 30 min intervals during 
the infusion. Thorough saline and air wash-outs of the bladder are 
used in terminating the collection of each sample. 

6) Plasma samples are obtained at 30^0 min intervals during 
the infusion. 

b) Calculation .—The total quantity of sucrose infused (2/F) at 
successive intervals is plotted against elapsed time. These points 
should fall on a straight line if the rate of infusion has been con¬ 
stant (see 2/F, Fig. 1). The total amount of sucrose excreted 
{ZUV) is calculated for each successive urine sample (see 2t/F, 
Fig. 1). These points should fall along a smooth curve and serve as a 
check on the accuracy of urine collection and analysis. The plasma 
concentration of sucrose, as determined at various intervals, is also 
plotted against time (P, Fig. 1). The virtual volume of distribu¬ 
tion of sucrose (F*) at any time during the infusion is obtained by 
inserting the appropriate interpolated values for 27F, 2C/F and P 
in the equation (F,, Fig. 1). 

The term P may be corrected for water content in this operation, 
but this involves superfluous labor and it is sufficient to make the 
preliminary calculations in terms of P as analytically observed and 
to apply the correction for plasma water only to the definitive figure 
for niaximal volume of distribution. This correction is effected by 
multiplying the final volume of distribution by the fraction of 
water in the plasma. 

At the rates of infusion used here (60.0-80.0 mg/min), the error 
involved in neg ecting blank excretion is less than 4 per cent of 27/F 

TV, uf*'at ‘he end of 120 min. 
he blank excretion is usually less than 1 per cent of 2/F-2t/F at 

the end of 60 min. Thus errors in F, are minimal even f the bLk 
excretion is omitted. 


Inuhn or sucrose may he determined by Roe’s resorcinol method 



as modified by Schreiner (31). This method depends on the reac¬ 
tion of resorcinol with the fructose liberated by hydrolysis of either 
substance. Plasma filtrates for inulin analysis are prepared by a 
cadmium sulfate precipitation (36). For sucrose, the plasma is pre¬ 
cipitated by Nelson’s (23) modification of the Somogyi zinc sulfate 
method. Urines are diluted directly unless protein is present, in 
which case they must be precipitated in the same manner as plasma 
before dilution. 

1) Preparation of plasma filtrate for inulin analysis. — a) Re¬ 
agents.—(1) Cadmium sulfate solution: 34.67 g 3 CdS 04-81120 
plus 169 cc IN H 2 SO 4 (exact) made up to 1,000 cc and filtered if 
necessary. (2) l.lN NaOH (exact). 

h) Procedure.—Two cc of plasma is placed in a 250 cc Erlen- 
meyer flask. Water, for the appropriate precipitation dilution, is 
added and mixed. Six cc of the cadmium sulfate solution is added 
with mixing. Two cc of NaOH is added dropwise with mixing. The 
flasks are stoppered, shaken well and allowed to stand for 10 min 
with some additional shaking. The solutions are centrifuged and 

the supernatant fluid is filtered through washed cotton. 

2) Preparation of plasma filtrate for sucrose analysis.—a) Re¬ 
agents.—(1) 5 per cent ZnS 04 - 6 H 20 . ’(2) Approximately 0.3N 
Ba(OH )2 (store in a bottle protected from CO 2 ). The ZnS 04 and 
Ba(OH )2 solutions are adjusted so that 4.7^.8 cc of Ba(OH )2 solu¬ 
tion will produce a definite pink to phenolphthalein with 5 cc of the 
ZnS 04 . The titration is carried out by diluting the ZnS 04 solution 
to 20 cc with water and adding the Ba(OH )2 with constant 

b) Procedure.—Two cc of plasma is placed in a 125 cc Erlen- 
meyer flask. Water, for the appropriate precipitation dilution, is 
added and mixed. Four cc of Ba(OH )2 solution is added slowly with 
mixing. Four cc of ZnS 04 solution is added slowly with mixing^ 
The flasks are stoppered, shaken well and allowed to stand for 5 
min. The solutions are centrifuged and the supernatant fluid 1 

filtered through washed cotton. 

3) Determination of sucrose or inulin in plasma filtrate or unm. 
a) Standards.—A series of standard solutions is run in duplicate 
with each group of unknown determinations. 

"ent.ato.lfor ?he sta..dard solutions are 1, 2 a..d 3 mg/100 cc for 

ce.d awfo'l't^re^ed d^ily). (2) 30 Pe^^Cl prepared by 

sta d!:r:rlution is pipetted into a 15 X 200 

Two cc of resorcinol solution .s added. cc ol .cage 



added. The tube is shaken and covered with a glass tear. The tubes 
are placed in a water bath at 80 C (± 1C) for 25 min and immedi¬ 
ately cooled in tap water for a minimum of 3 min. 

Color intensity is read in a Coleman Junior spectrophotometer 
with matched cuvets at a wavelength of 490 m^ using an aqueous 
reagent blank for 100 per cent transmission. This measurement 
should be carried out promptly, although the color developed is 
reasonably stable for over an hour. 

Critique of methods. —The calibrated infusion technique is a 
more satisfactory approach to the measurement of extracellular 
fluid than the constant infusion-total urine collection method. 
Absence of the requirement for lengthy and complete urine collec¬ 
tion permits application of the method to children and to most 
adults. If sucrose is used, a valid determination is usually completed 
within 180 min. That uniform distribution throughout the extra¬ 
cellular fluid has been achieved in each determination is revealed 
by the attainment of a constant volume of distribution. Evidence 
that maximal volume of distribution has been achieved in the con¬ 
stant infusion-total urine collection method can only be obtained 
by repeating the procedure over a longer period in the same sub¬ 

If facilities are not available for administration of an infusion of 
the requisite constancy, the calibrated infusion technique cannot 
be used successfully and the constant infusion-total urine collection 
method should be used. Since it may be anticipated that plasma- 
interstitial fluid equilibrium is obtained more rapidly with sucrose 
than with inulin, it is recommended for use in man by either method. 
Ony mulin should be used for measurement of extracellular fluid 
in dogs, since sucrose is metabolized in this species (26). 

Measurement of increased extracellular fluid volume.~W\ih. 
ceitain modifications, the calibrated infusion technique may be 
used to measure increased volumes of extracellular fluid (4). A 
longer period of infusion is required in all such instances. 

In subjects with edema but without sufficient accumulation of 
pleuial or peritoneal fluid to permit convenient sampling of these 

DL's1bl^\f It continued for 8 hr 

possible. Results are analyzed in the manner previously de¬ 
scribed. Samples of edema fluid should be collected before the in- 
fusion IS interrupted as additional evidence that plasma-edema fluid 
eiiuihbrium has been accomplished. 

In edematous subjects with accumulations of ascitic and/or 
|) eural fluid the infusion should be continued as long as possible 



peritoneal fluids should be obtained during the infusion and used 
to define the time course of sucrose concentration in the extracellu¬ 
lar depots and demonstrate equilibrium distribution. 

hepatic cirrhosis. 

If maximal volume of distribution has not 
concentrations may permit ^ 

fluid volume by application of the P^'^'P u bv dS 

•i.rr. tViP PYtracellular fluid volume may be calculated by S 

rium, the extraceiiuiai uu (y'jV-'^UV) by sucrose concen- 

the amount of sucrose in the body v) vy 



tration in plasma water, pleural fluid, ascitic fluid or edema fluid. 
Before equilibrium distribution, all volumes of distribution are 
virtual and, when calculated from the plasma concentration, in¬ 
crease to the limiting value of the maximal volume of distribution. 
However, the virtual volume of distribution, calculated from the 
concentration in the extracellular fluid depot which is slowest to 
equilibrate, is initially infinitely large and decreases to the volume 
calculated from the plasma concentration. 

The virtual volumes of distribution based on plasma and extra¬ 
cellular depot (pleural, ascitic or edema fluid) concentration are 
plotted against elapsed time (see F? and Vf, Fig. 2). If these esti¬ 
mates are within 10 per cent of one another at the time the infusion 
is discontinued, the average volume or the volume obtained by 
extrapolation is a good estimate of extracellular fluid volume, 
despite the fact that uniform distribution has not been attained. 

Interpretation of dilution (space) techniques using radiosodium, 
radiochloride and bromide. —It is now recognized that sodium and 
chloride are normally present in certain cells and that estimations of 
extracellular fluid volume by dilution of these substances are not 
valid. Indeed, the use of volume units and the term “space” in 
reporting such dilution studies tends to obscure the important 
information which their use should disclose. 

The fundamental concepts of tracer theory require that isotopic 
Na and Cl be regarded as diluted in body Na and Cl and not in 
body water. In reality, the measurement is that of a mass and not a 
space. It is assumed that when the isotope has reached maximal 
dilution the ratio of the concentration of isotope (counts/min) to 
that of the nonisotopic form (mEq) t is the same in all tissues of 
the body regardless of water content. 

After corrections for radioactive decay, the total body elec¬ 
trolyte may be calculated as 

Total body isotope in- ”1 _ T amt of isotope ex- "1 

electrolyte = Ljected (counts/min)J _Lcreted (counts/min )J 

(mEq) p isotope concen. in plasma ( counts/min 

L electrolyte concen. in plasma (mEq/cc) J 

As may be seen from the formula, correction must be made for 

isotope during the time of complete dis- 

It has been demonstrated that distribution of administered bro¬ 
mide IS similar to that of chloride in tissues and body fluids except 

chlorX mavTe Consequently, total body 

^niond^ay be estimated by bromide dilution (8, 40). When 

t This is termed the specific activity (counts /min/mEq). 


complete distribution of bromide is effected, the calculation is 

t aint of Br“ in-"| _ famt of Br~ ex-"] 
_ jected (mg) J _L creted (mg ) J 

L Br~ concen. in plasma (mg/cc)"] 

Cl concen. in plasma (mEq/cc)J 

Simultaneous measurements of total body electrolyte (Na or Cl) 
and body water distribution permit calculation of intracellular 
electrolyte and intracellular electrolyte concentration. This prin¬ 
ciple has bee'll applied to dogs (19) and man (6, 7). 

It is not possible at present to state with certainty the extent to 
which so-called “bound” electrolyte exists in man. “Bound” elec¬ 
trolyte, presumably in bone and possibly in other tissues, would 
not be measured by these techniques. In the absence of definitive 
information on this point, the term “exchangeable” (2) has been 
proposed as descriptive of the physiologic state of the total elec¬ 
trolyte mass as measured by the isotope dilution technique. It 
seems appropriate, then, to refer to the values obtained as total 
exchangeable Na or Cl, etc. 

A minimal equilibration time of 24 hr for sodium, bromide and 
chloride is recommended for the most accurate measurements. 


1. Isotope techniques—The use of deuterium oxide (P 2 O) for 
measurement of total body water in man was initiated by Hevesy 
and Hofer (15). Pace and his co-workers (25) have described the 
use of tritiated water (water labeled with tritium [HTO], the radio¬ 
active isotope of hydrogen, with mass 3) for this measurement. On 
theoretical grounds these substances seem to be ideal. In tracer 
(luantities, the biologic behavior of “tagged” water is not signi - 
cantly different from that of the nonIsotopic form. The greatest 
volume of dilution has been shown to agree f'^osely with measure¬ 
ments of total body water obtained by desiccation (21, 22, 25). 


1) The substance is administered by a single iiijection wit a 
calibrated syringe or buret. From 50 to 100 cc of D 2 O is the usual 
dose; larger amounts are required if total body water is much m- 
ereaid In general, the precision of the method improves with the 
use of larger amounts, since the higher equilibrium DjO conceiitia- 
tion will be in a range of increased analytic accuracy from 0*1 to 
equilibrium concentration with these amounts will be f'om ai to 
0^2 per cent D^O. The dose of tritiated water may range fio . 
to 25 0 me which yields equilibrium concentrations o 0.1-0.- 
me/l If l^rial sOidieLire being coiiducteil. a blood sample should 



he obtained before injection for determination of deuterium or 

tritium blank. 

2) In the absence of abnormal increments of body water, equilib¬ 
rium distribution will be attained in 2 hr. At least 2 blood samples 
should be collected between the 2d and 4th hours to demonstrate 
that equilibrium has been reached. With increased total body 
water and abnormal extracellular depots, an interval of 4-6 hr 
should elapse before the 1st sample is obtained and additional blood 
specimens should be collected to the 12th hour. In such instances 
samples of ascitic, pleural or edema fluid should be obtained at the 
time of collection of the last blood sample as additional demonstra¬ 
tion that homogeneous distribution has been attained. 

3) If it is anticipated that the time required for equilibration will 
be in excess of 4 hr, all urine should be collected from the time of 
injection. Otherwise it is sufficient to estimate the loss during the 
first 2 hr through all routes as 0.4 per cent of the amount injected 

Analysis of deuterium and tritium .—Limitations of space pre¬ 
vent a complete description of the methodology pertaining to deu¬ 
terium and tritium analyses. 

Deuterium concentration may be determined directly, after con¬ 
version of the sample to a gas, in the mass spectrometer (for a 
general description of this instrument, see (20)). Detailed informa¬ 
tion concerning types of instruments and applications of mass 
spectrometric methods has been presented by Siri (34) and Nier 
(24), and Solomon, Edelman and Soloway (38) have presented 
specific considerations of the mass spectrometric analysis of 
deuterium in body fluids. 

The concentration of DqO in water may be determined indirectly 
by measuring certain physical properties of deuterium oxide-water 
mixtures which vary with concentration of the former. Robertson 
(27) has described the indirect methods for measuring D 2 O with 
particular reference to methods based on density differences (fall¬ 
ing-drop, free-float, pycnometer and diffusion gradient methods) 
and chanps in the refractive index. Of these methods, the falling- 
drop method using o-fluorotoluene as initiated by Keston, Ritten- 

berg and Schoenheimer (17) has been widely used in biologic investi¬ 

that tritium is radioactive permits determination of 
emitted ladiation. Tritium is a /^-emitter with a half-life of 10.7 yr 
4 he radiation is of extremely low energy (0.011 mev) so that 
spec,a .ech,jiques must be applied. Pace^nd his coTorkens 25 
a ye (^v,.sed a counting assembly which involves the introduction 
»l m O as water vapor at low pressure into a Geige,- tube, the emis" 



sion being counted directly. However, the sample may be con¬ 
verted to the gas phase before introduction into the tube. Eidinoff 
and Knoll (10) have compared gas counting with a method for 
measuring tritium in solid samples. The vibrating reed electrome¬ 
ter has also been adapted for tritium analysis (35). 

Calculation .—Total body water is calculated by dividing the 
total amount of “tagged” water administered (after correction for 
urinary excretion) by the equilibrium concentration in plasma 


2. Antipyrine method .—The use of antipyrine in measurement of 
total body water, proposed by Soberman and co-workers (37), has 
been described elsewhere in this series (1), as has also the use of 
N-acetyl 4-amino antipyrine for this purpose (1). Descriptions of 
these methods are therefore omitted here. 

Critique of methods .^—Except for the error introduced by the ex¬ 
change of the isotope with hydrogen atoms of organic molecules, 
D 2 O and HTO are ideal substances for the measurement of total 
body water because of their rapid equilibration and successful 
biologic masquerade as water. For D 2 O, the error through exchange 
has been estimated to produce results which are high by a factor 
of 1.0-1.5 per cent of body weight (700-1,000 cc) (38). Unsatisfac- 
torv studies with these materials can only result if insufficient time 
is allowed for homogeneous distribution or if there are technical 
errors in the administration of the infusion or collection and analy¬ 
sis of the blood samples. The methods have the disadvantage that 
the analytical techniques are complex and require special equip¬ 
ment and trained personnel. It should be pointed out that there is a 
fairly considerable change, percentagewise, m the molecffiar weight 
of these substances as compared to water. The ratio of the molecu¬ 
lar weights of D 2 O and HTO to water is 1.11. Hence ^herj is a 
possibility that their rates of internal exchange m the body m y 

differ from those of nonisotopic water. ,, , 1 

The antipyrine method is a valuable contribution to methodology 
inThis field because of the simplicity of i 

proce’)lu,-e be modified if‘otal body water 
increased The 1st blood sample should not be taken be 
and the additional samples should beobtained for as long as po^sib , 
■ • Tr> r»f 12 hr If there is a pronounced increase in ext 


Theoretically, the lack of equilibration should be revealed by a 
changing slope of the semilogarithmic plot; actually, the rate of 
equilibration in certain slowly exchanging depots may be com¬ 
parable to the rate of metabolism, in which case the graph will not 
separate these components. 

The 2d reason derives from the fact that, for a valid measure¬ 
ment, the antipyrine concentration must demonstrably be the 
same in all parts of extracellular fluid. If differences of concentra¬ 
tion or slope occur among components of extracellular fluid, this 
condition has not been fulfilled. This point requires emphasis be¬ 
cause in some subjects with conspicuously increased extracellular 
fluid the concentration of antipyrine in pleural or ascitic fluid may 
exceed that in the plasma water, a discrepancy that may persist 
despite a subsequent decrease in both plasma and extracellular 
depot concentrations as a result of metabolism (4). The error leads 
to an overestimate of total body water. 


The only method available for the estimation of intracellular 
water is indirect; i.e., intracellular water must be calculated as the 
difference between total body water and extracellular fluid. The 
methods for measuring these 2 compartments have been described 
in detail. Obviously the accuracy of the derived value for intracel¬ 
lular water will be determined by the individual errors in the pri¬ 
mary methods. 

Normal and abnormal patterns of body water distribution can be 
effectively expressed if intracellular water is calculated as per cent 
ot total body water and also per cent of cellular mass (body weight 
minus weight of the extracellular fluid) as well as per cent of body 
weight. In 11 normal males, intracellular water averaged 70.2 per 
cent of body water (range 64.9-74.5), 50.3 per cent of cellular (41.8-59.6) and 40.6 per cent of body weight (range 34.7- 
48.0) (3). Changes in these terms readily show shifts of fluid and 

alterations in cellular hydration (intracellular edema or dehydra¬ 

Comment by David P. Earle, Jr. 

Although time-consuming, the methods described here represent consid¬ 
erable improvements m the measurement of body fluid compartments 

thiocyanate' Ind" the T distribution of 

tmocjanate and the like. However, extreme care must be devoted to all 

technical details if accurate results are to be obtained. Further it must be 

kept mmd t mt any “apace” measured by the distribS onnulin 

sucrose and antipyrine is tramslated to extracellular fluid m- tntoi k a ’ 

water by inference. Nevertheless, these credit; “b:st‘?a™i,abt 



Tlie calibrated infusion technique for the measurement of extracellular 
fluid is particularly attractive because of the short time required and be¬ 
cause of the internal check available when a constant volume of distribu¬ 
tion is achieved. However, if uniform distribution of sucrose in edematous 
subjects is not achieved in all parts of extracellular fluid (plasma, intersti¬ 
tial, pleural or ascitic fluid) at the same rate, the interpretation of results 
is open to question. The alternative method of measuring sucrose in all 
available fluids is tedious. In the case of inulin, a constant infusion of as 
long as 24 hr or more may be required before the concentration of inulin 
in pleural or ascitic fluid becomes equal to that in plasma. 

\\ hen the resorcinol method, which does not involve the use of yeast to 
destroy glucose or fructose in the blood, is used to measure inulin or su¬ 
crose it is essential that the subject be in the fasting state. The edematous 
subject may therefore be required to go without food for almost 24 hr in 
view of the long time required to achieve uniform distribution. 

Comni-ent by Isidore S. Edelvian 

Dr. Deane is to be congratulated on covering the various techniques in 
such concise and useful form. The following remarks may be pertinent. 

1. There is some evidence that thiosulfate may prove to be as accurate 
and more convenient in measurement of extracellular fluid space than 
either inulin or sucrose. 

2. It is true, as Moore (30) pointed out, that the exchangeable sodium 
(mEq) is the important parameter measurable with radiosodium. Never¬ 
theless, the sodium space (liters) may still be a useful and perhaps even a 

nhvsiologicallv important value. . 

3 In the ekimate of exchangeable sodium, some 600-900 mEq is con¬ 
tributed bv bone sodium (12, 16). Since the penetration of substances for 
measurement of extracellular fluid into bone has not been re|»rted, we do 
not know bow much of this sodium is included m estimates of mtracelluiai 

sodium by isotopic dilution methods. * +■ 

' 4. A revised and accurate technique for measuring DoO concentrations 

bv the falling-drop method has been described (30). 

^5 in Imv eoncLtrations, as is the case when D,0 is used as a water 
tracer, almost all of the tracer exists as HDO. The molecular ratio to «a 
hpinp- 1 05 the likelihood of a “mass effect is small. 

r D O ha^s been applied to studies on the rates of water exchange be- 

Si^fne— C“ slii'rf:::: 

utf Jibe artiilal curve in terms of transcapillary water exchange is 

Comment by Dr. Deane 




|K)stinfusion decrement in plasma concentration. A rough calculation indi¬ 
cates that the rates of diffusion of sucrose and thiosulfate would be about 
the same; the comparative advantage of either would presumably depend 
on the ease and accuracy of chemical methods. 

2. The comment that the concept of “sodium space (liters) may still be 
a useful and perhaps even a physiologically important value” does not ac¬ 
cord with Dr. Edelman’s recognition and acceptance of the fact that the 
function measured is exchangeable sodium (niEq). 


1. Brodie, B. B.: Measurement of Total Body Water, in Visscher, M. B. 

(ed.): Methods in Medical Research (Chicago: Year Book Publishers, 
1951), Vol. 4, p. 31. 

2. Corsa, L., Jr., ei al .: The measurement of exchangeable potassium in man 

by isotope dilution, J. Clin. Invest. 29: 1280, 1950. 

3. Deane, N.: Intracellular water in man, J. Clin. Invest. 30: 1469, 1951. 

4. Deane, N.: Unpublished observations. 

5. Deane, N.; Schreiner, G. E., and Robertson, J. S.: The velocity of dis¬ 

tribution of sucrose between plasma and interstitial fluid, with refer¬ 
ence to the use of sucrose for the measurement of extracellular fluid in 
man, J. Clin. Invest. 30: 1463, 1951. 

6. Deane, N., and Smith, H. W.: The distribution of sodium and potassium 

in man, J. Clin. Invest. 31: 197, 1952. 

7. Deane, N.; Ziff, M., and Smith, H. W.: The distribution of total body 

chloride, J. Clin. Invest. 31: 200, 1952. 

8. Dunning, M. F.; Steele, J. M., and Buerger, E. Y.: The measurement of 

total body chloride, Proc. Soc. Exper. Biol. & Med. 77: 854, 1951. 

9. Edelman, I. S., and Moore, F. D.: J. Clin. Invest. 30: 637, 1951. 

10. Eidinoff, M. L., and Knoll, J. E.: The measurement of radioactive hydro¬ 
gen in solid samples—comparison with gas counting. Science 112: 250, 

11 . 

12 . 







20 . 

Flexner, L. B., et al: Cold Spring Harbor Symposia 13: 88, 1948. 

Forbes, G. B., and Perley, A.: Estimation of total body sodium by 
isotopic dilution: I. Studies on young adults, J. Clin. Invest. 30: 558 
1951. ’ 

Gaudmo, M., and Levitt, M. F.: Inulin space as a measure of extracellular 
fluid. Am. J. Physiol. 157: 387, 1949. 

Gaudino, M.; Schwartz, I. L., and Levitt, M. F.: Inulin volume of dis¬ 
tribution as a meiusure of extracellular fluid in dog and man, Proc Soc 
Exper. Biol. & Med. 68: 507, 1948. 

Hevesy, G and Hofer, G.: Elimination of water from the human body 
Nature, London 134: 879, 1934. 

JS'f’ \ ’V preparation, 

eston, A. S., R.ttenberg, D., and Schoenheimer, R.: Determination of 

T I organic compounds, J. Biol. Chem. 122: 227 1937 

Lentt, M F., and Gaudino, M.: Use of radioactive isotopes to measure. 
15 »“ot, inir'™" i" ‘he normal dog, Am. J. Physiol. 

Lill.v, J. C.: Physical Methods of Respiratory Gas Analysis, in Comroe, 


21 . 

22 . 





















J. 11., Jr. (ed.); Methods in Medical Research (Chicago: Year Book Pub¬ 
lishers, Inc., 1950), Vol. 2, p. 131. 

McDougall, E. J., et al.: Heavy water in the animal body. Nature, London 
134; 1006, 1934. 

Moore, F. D.: Determination of total body water and solids with iso¬ 
topes, Science 104; 157, 1946. 

Nelson, N.: Photometric adaptation of the Somogyi method for determi¬ 
nation of glucose, J. Biol. Chem. 153; 375, 1944. 

Nier, A. O.; in Glasser, O. (ed.); Medical Physics (Chicago; Year Book 
Publishers, Inc., 1950), Vol. II, p. 474. 

Pace, N., et al.: Studies on body composition: IV. Use of radioactive hy¬ 
drogen for measurement in vivo of total body water, J. Biol. Chem. 168: 
459, 1947. 

Power, M. H., and Keith, N. M.; Experiments on distribution and renal 
excretion of sucrose injected intravenously in dogs, J. Biol. Chem. 114: 
Ixxx, 1936. 

Robertson, J. S.: in Siri (34), p. 263. 

Schachter, D.; Freinkel, N., and Schwartz, I. L.; Movement of inulin 
between plasma and interstitial fluid. Am. J. Physiol. 160: 532, 

1950. , . ^ 

Schloerb, P. R., et al.: The measurement of total body water in the 
human subject by deuterium oxide dilution, J. Clin. Invest. 29. 1296, 

1950. . . , . u J « -J 

Schloerb, P. R., et al.: Measurement of deuterium oxide in body fluids 

by the falling drop method, J. Lab. & Clin. Med. 37: 653, 1951. 
Schreiner, G. E.; Determination of inulin by means of resorcinol, Proc. 

Soc. Exper. Biol. & Med. 74: 117, 1950. 

Schwartz, 1. L.; Schachter, D., and Freinkel, N.: The measurement of 
extracellular fluid in man by means of a constant infusion technique, 
J. Clin. Invest. 28: 1117, 1949. 

Schwartz, 1. L.: Measurement of extracellular fluid by means of <)onstant 
infusion technique without collection of urine. Am. J. Physiol. 160. 

Sirl^W. E^-!/so/opfc Tracers and Nuclear Radiations (New York: McGraw- 
Hill Book Company, Inc., 1949), p. 279. 

Siri W E , and Robertson, J. S.; Unpublished observations. 

Smith H W., et al.: The renal clearances of substituted hippuric acid 
derivatives Ind other aromatic acids in dog and man, J. Clin. Invest. 

80 ^ 2 , R.! al.: The use of antipjTine in the measurement of total 

SoSn:Ty ■ Sn:an’:7'£^;:i ^ .p- use otmasssp. 

trometer to measure deuterium in body fluids, J. Uiii. Invest. 

311, 1950^ .„aBrodie B B.; The distribution of administered bro- 

llr'in°co"m;a"* chloride and it, elation to body fluids, J. 
■harmacol. & animals fol- 

i^!,®J,^“s‘::dirCmte, Am. d. Physiol. 127: 33S. 

if E G., and Hastings, A. B.i The 

idein tissues and body fluids, J. Biol. Chem. 129: 547, 1939. 


III. Electrolyte Excretion Studies in the Dog 

LAURENCE G. WESSON, JR., New York University-Bellevue Medical Center 

The techniques described here are aimed at evaluating the rela¬ 
tive contributions to renal electrolyte and water excretion of plasma 
composition, glomerular filtration rate and tubular function. To 
this end it is essential that composition of the plasma be main¬ 
tained at constant controllable levels. Two situations specifically 
described which illustrate these principles are the effects of expan¬ 
sion of extracellular fluid (p. 181) and urea diuresis (p. 188). 


All experiments are performed on postabsorptive female mongrel 
dogs accustomed to the procedures. No food is given them for 18 
hr. We allow water ad lib but avoid forced hydration the night be¬ 
fore an experiment because, in man at least (6), it leaves trace 
effects on the responses of sodium and water excretion to intrave¬ 
nous saline infusion 12 hr later. 

Glomerular filtration rate is measured as the creatinine clearance 
by the technique of urine collection, and simultaneous plasma deter¬ 
minations* which permits estimates of rapid changes in glomerular 
filtration rate, as other methods do not (12). Effective renal plasma 
flow is measured by the clearance of p-aminohippurate at low plasma 
levels (12). 

Sodium and potassium are measured by the flame photometer 
against an internal lithium standard (3). Chloride is determined by 
the method of Sendroy as modified by Van Slyke and Hiller (14). 
Plasma and urine carbonate are computed (9) from total CO 2 and 
pH, using the HeiKlerson-Hasselbach equation. 

^ Plasma proteins are determined by the copper sulfate method of 
Van Slyke et al. (15). For the dog we use the factor (369) and 
specific gravity of plasma water (1.0065) described for man, al¬ 
though true dog values are probably somewhat different. When 
plasma sodium and urea concentration are greatly altered, correc¬ 
tion factors are used. The general equation we use is 

P == 369(S - 1.0065 - Ky - K^) 

^ 1 V • '*****^- 11V./VYO 111 cAi;t;sa 01 o mi/min t 

the lower the 

le bladder is not rinsed. Since 
urine flow, the longer should 

be the collection periods. 



should be substituted in experiments involving other solutes. 

The freezing point of plasma and of urine is measured with a 
thermistor in a Johlin freezing point apparatus (5). The freezing 
point of biologic fluids can thus be measured as accurately as with a 
Beckman thermometer but more rapidly and with smaller samples. 

THERMISTOR. —A tliermistoi’ (Western Electric type 14B) con¬ 
sists of a small bead of copper oxides which displays in the region of 
0 C an increase in resistance of 250-350 ohm/degree decrease in 
temperature. It is mounted about Vie in. from the tip of a long 
medicinal applicator stick. Fine, bare copper wires are soldered to 
the terminals of the thermistor and run up the stick, where they 
are soldered to heavier wire leads. This assembly is given 5 coats 
of waterproof cement, each coat being oven-dried at 100 C for at 
least 24 hr and the whole assembly, beginning 1 cm above the ther¬ 
mistor bead, is given a single tight wrapping of electrical Scotch 
tape. The thermistor assembly is incorporated into a standard 
Wheatstone bridge circuit, (e.g, Leeds and Northrup no. 4760). A 
1.5 V dry cell battery serves as current source. Balance is measured 
with a galvanometer of medium sensitivity (0.01 n amp/mm). A 
1,000 ohm shunt with a snap switch is placed across the poles of the 
galv'anometer for coarse adjustment. 

FREEZING POINT DETERMINATION. —^The freezing apparatus and 
procedure are similar to those described by Johlin (5). Samples, 
0.5-2.0 ml, are placed in long plastic tubes about 1 cm in diameter 
(our tubes are made by cementing together portions of shorter 
tubes). After precooling in a cracked ice bath, a sample tube is 
placed in the freezing apparatus and the thermistor assemb^ 
placed in the tube. The temperature of the freezing mixirire is ad¬ 
justed to about 2° C below the expected freezing point of the sample 
and the sample cooled to about 0.5° C below its expected freezing 
point. At this temperature, about 1 per cent of the water is con- 
\'erted to ice on abolition of supercooling. With the galvanometer 
shunt in the circuit, and with continuous stirring by a gentle up- 
down motion of the thermistor assembly, the progress o cooling 
may be followed by the increasing resistance necessary to main¬ 
tain a null reading on the galvanometer. For best results, theie 
should be continuous current flow, rather than make-break, hen 
the region of the freezing point is reached. Supercooling is abolis e 
with a fro.sted 2 mm section of lu^axT Suuge copper wne 1 ns is 
inai-ked hv a rapid decrease in resistance. A plateau is reached in a 

lew seecm;isuiKl lieW lor s.-ve.ul mmiites. &,al S' 

0,1 the Kulvaiionieter shunt is opened and tJie final adjust 
of the resistance made. For practical purposes, the resistance 

ment of 

ortliis plateau, read to the nearest 0.5 ohm, may be taken as meas 


uring the freezing point. Thermal equilibrium is reached very 

Actual freezing point is obtained from a standard curve pie- 
pared by plotting the freezing point of a graded series of standard 
solutions (e.g., urea, glucose, Nad) against their plateau resist¬ 
ances. In practice, we take the average of 3 separate determina¬ 
tions, which should not differ, one from another, by more than 2 
ohms. In contrast to the Beckman thermometer, the thermistor 
resistance element is a continuous source of heat. A gradient of heat 
flow is established between thermistor and the freezing bath so that 
the measured temperature is the warmer end of a system in dy¬ 
namic equilibrium. Accuracy depends primarily on careful tempera¬ 
ture control and duplication of conditions between standard and 
unknown solutions. With practice, the freezing point of 4 samples/- 
hr (3 determinations/sample) <*an be analyzed with an accuracy of 
0.003° C. The following additional notes on use of this procedure 
should be useful. The standard curve shifts within a narrow range 
and should be checked every few hours. We believe that shifting of 
the standard cur\'e is due largely, if not entirely, to changes in 
loom temperature and humidity. The thermistor assembly should 
be thoroughly dried when not in use; water penetration with short- 
circuiting has been our major source of trouble. The battery should 
be checked eveiy 2 weeks; a weakening battery gives poor checks 
on duplicate determinations. 

Because of the very low heat content of the system compared 
with the sample (a major factor in permitting ease and accuracy of 
freezing point measurement), we believe this apparatus can be 
easily modified to measure the freezing point of samples of 0.1 cc or 

osmolaritv = point depression of sample 

(-)1.86° C 

Osmotic piessure of plasma and urine may be estimated from 
their composition provided all constituents are known. In certain 
circumstances a few known substances comprise most of the solute 

approximated. Thus, under urea loading, 
99 per cent of the solute may consist of urea plus salts of sodium 
and potassium. Approximately 95 per cent of the osmotically active 
solute of plasma IS salts of sodium and potassium plus urea and In ur(*a diuresis w(‘ use the general formula 

osmolality = [meaj + [gluoo.sej -f- 2(lNa] + [K]) X F 

where F represents tlie deviation of electrolytes from the freezing 
pomt caused by electolytes in contrast to substances such as urea 



and glucose. For practical purposes, total electrolyte may be con¬ 
sidered similar to NaCl, the effect of which on freezing point at 
different concentrations is obtainable from standard texts, e.g., 
International Critical Tables. F varies in a nonlinear manner from 
1.0 at zero electrolyte concentration to 0.90 at 500 mEq/1. 

Calculations .—The rate of reabsorption of a substance from the 
glomerular filtrate is given by the general equation 

T = {Cf X P) -{U XV) 

where T is the rate of reabsorption (mlVI/min), Cf is glomerular 
filtration rate (cc/min), P is plasma concentration of the sub¬ 
stance, and Cf X P is the filtered load (mM/min). U is concentra¬ 
tion in the urine (mM/ml), V the urine flow (cc/min), and U X V 
is the urinary excretion rate (mM/min). 

The minute excretion rate of an ion being simply computed 
{U X V), computation of T depends on the values obtained for 
filtered load (Cf X P)- Since the glomerular filti-ate has essentially 
the composition of a simple plasma ultrafiltrate, 2 corrections must 
be applied to the term Cf X P) one is for the fraction, w, of filtrable 
volume in the plasma, and the other, k, for the inequalities of dis¬ 
tribution of ions produced by restriction to the vascular compait- 
ment of the negatively charged protein molecules (Donnan effect). 
No significant inaccuracy is involved in the assumption that the 
weight of plasma proteins is a reasonably close measure of the 
volume of nonfiltrable plasma materials, so that w - Pprot 
Donnan factors are obtained from the literature (2, 4) by plotting 
plasma protein concentration against corresponding ratios ol extra- 
vascular over vascular ion concentrations as measured kdo- 
KTsim of water. The best straight line is then drawn through the 
point of protein concentration = 0 and ion ratio = 1. 1 he data do 

liot estaWish the relationship as truly linear and 

probably be refined. For a protein concentration difteieiice of 6 g 

per cent between plasma and ™7 e 

values of 1.03 for Cl' and HCO,-, 0.95 for Na+ and 0^89 for K . 
'I'he rate of formation of glomerular filtrate is iiow wCf, and t 
conceutraihui of any ion in the filtrate is kP/.. The tubular load, 
after cancellation of terms, is Cf X kP. 


It has lonir been recognized that the kidney plays a major role in 

t For a review of the literature of this field, see ( 13 ). 



Tlie kidney’s role in regulating the volume of the extracellular fluid 
is less clear and the mechanisms involved are obscure. 

In man, when extracellular fluid volume is acutely expanded, as 
by injection of Locke’s solution, excretion of salt and water in¬ 
creases in a few minutes and reaches a maximum m a few hour^ 
The urine may attain a flow of 5-10 cc/min and the inuhn U/E 
ratio may fall'in the range 15-20. If the extracellular fluid volume 
of a dog is acutely expanded, increase of excretion of salt and water 
begins more rapidly than m man and reaches a maximum within an 
hour, with creatinine U/P ratios of 5-10. High, complexly varying 
rates of salt and water excretion continue until the amounts elimi¬ 
nated equal or slightly exceed those injected. Concomitant effects of 
loading are increases in central venous pressure and dilution of 
plasma proteins; the former does not prevent the increase in salt 
output, nor does the latter account for the increase in filtration 
rate observed in the dog. 

In contrast, simple depletion of extracellular fluid volume in man 
or dog depresses outputs of salt and water, so that the urine may 
contain no more than traces of sodium, while w'ater loads are slowly 
excreted. In this circumstance the body tends to restore extra¬ 
cellular fluid volume by diluting that which remains wdth water, so 
that plasma sodium concentration is commonly decreased. In spite 
of the fact that the plasma becomes hypotonic, water diuresis 

These responses to repletion and depletion of extracellular fluid 
pose 2 problems: (1) what are the changes in renal function w^hich 
result in retention or excretion of salt and water, and (2) what 
organ or tissue responds to changes in extracellular fluid volume 
and signals for an appropriate adjustment of renal function? 

Changes in renal excretion of salt result from changes either in 
filtered load or in tubular reabsorptive function. In the dog, filtered 
load, as influenced by changes in filtration rate,.seems to be the 
more important control of salt excretion. Administration of Locke’s 
solution or feeding of large amounts of NaCl may increase filtration 
rate by as much as 50 or 75 per cent. Almost certainly as a result of 
this, there occurs a massive increase in excretion of salt and w'ater, 
so that saline equal in quantity to as much as half the normal extra¬ 
cellular fluid volume may be excreted over 2-4 hr; filtration rate 
returns to the control level as elimination of the load becomes com¬ 
plete. Participation of changes in tubular reabsorption in this re¬ 
sponse has not been fully evaluated. During acute loading, the 
amount of salt reabsorbed increases, whereas it might have been 
expected to decrease; how^ever, this increase in reabsorption is more 
than compensated for by the increase in filtration. Whether or not a 



decrease in reabsorption ensues when the loading is sustained for 
more than 24 hr is not known. 

Depletion of the extracellular fluid in the dog causes a drop in 
filtration rate sufficient to account for the ensuing retention of salt 
water; whether or not tubular reabsorption is also increased is not 
known. Thus, in the dog, changes in glomerular filtration rate are 
probably the primary mechanism in regulating extracellular fluid 

Man does not exhibit a comparable flexibility of filtration rate. 
The filtration rate of a human subject responds to a load of Locke’s 
solution by no change or by increases of only 10-15 per cent.J At 
the same time, excretion of a load of Locke’s solution is much more 
sluggish and may require 24-28 hr. Although man is unable to draw 
freely on filtration rate in control of salt excretion, tubular reab- 
sorptive capacity under conditions of saline loading or a high salt 
diet is believed to change in the direction of a slight but definite 
decrease, so that salt excretion increases. Salt depletion is accom¬ 
panied by a decrease in filtration rate which is not as pronounced 
as in the dog. The extent to which tubular reabsorptive capacity 
may increase in salt depletion in man is not clear. 

The end-organs responsive to change in extracellular fluid volume 
and the pathways by which they evoke the appropriate changes in 
renal function are unknown. Some scant evidence suggests that the 
volume-sensitive organ is at least partially integrated with the 
osmoreceptor apparatus of the neurohypophysis. Thus, Ladd (6) 
has shown that water diuresis conspicuously alters the kidney s re¬ 
sponse to a saline load infused some hours later, so that both salt 
and water excretions are enhanced. Nelson and W elt (8) have 
adduced evidence that water retention and creation of hypotonic 
plasma in cirrhotics may be due to enhanced antidiuretic hormone 

The procedure that follows was devised to study renal function 
under conditions such that volume and composition of the ex¬ 
panded extracellular fluid are constant during the oxpenmenL 
Ideally we should like to fix all conditions and examine the result 


of variables subject to control by the investigaioi. 

t In certain circumstances, 

probably contribute extensively to uigg filtration rate: a sharp rise in fil- 




The object of this experiment was examination of the effects on 
renal function of an increase in body salt and water without change 

in composition. 

a) Solutions—A: Creatinine (Cr) and PAH prime. Per kilogram 
of body weight: Cr, 50 mg; PAH, 8 mg; Locke’s solution (same as 
saline prime, C below), 1.5 ml. The prime will produce a plasma Cr 
concentration of about 12 and PAH concentration of about 2.0 mg 
per cent. 

B: Cr and PAH sustaining solution. Composition is computed 
from estimated excretion rate (P X C, the latter term determined 
by experience) divided by pump infusion rate. Rate of infusion 
should be low, less than 1.0 cc/min, to avoid introducing more 
fluid than necessary by this route. Cr and PAH may be dissolved 
in distilled water or saline, as desired; since both are buffers, solu¬ 
tions containing them should be titrated to pH 7.4 with strong 
acid or base. 

C: Saline prime. We use the following formula based on mean 
calculated plasma ultrafiltrate composition of our animals. NaCl, 
124 mEq/1; KCl, 5 mEq/1; NaHCOa, 26 mEq/1; CaCL, 1 mEq/1; 
glucose, 100 mg/100 cc; creatinine, 12 mg/100 cc; PAH, 2.0 mg/- 
100 cc. The solutions, so far as possible, should be prepared Avith 
pyrogen-free water, and all glassware and tubing should be chemi¬ 
cally clean. 

D: Saline solution. This is a series of solutions designed to ap¬ 
proximate clo.sely the electrolyte composition of the copious urines 
which may be excreted during an experiment; the appropriate 
solution is infused into the animal in a volume equal to the urine 
flow. The composition of the urine may be computed by rapid 
analysis during an experiment or predicted from analvsis in pre¬ 
liminary experiments. 

h) Technique—The animal is placed on a comfortable dog 
boaid. An indwelling plastic catheter (7) about 5 in. long is inserted 
into each external jugular vein through a thin-walled 14 gauge 
needle. The needle is wnthdrawn and an adapter consisting of a 
truncated 19 or 20 gauge needle is forced into the catheter lumen 
4 he catli^eters are fliisluHl with a drop of heparin solution (Liquae- 
mni, Hoflmaiin-La Roclie), stopp(‘r(Hl by any convenient means and 
fastened m phu^ with a, bit of adhesive and surgical skin clips or bv 
stitch tin (High the skin. ()ne cathet er is for inject ion of solutions 
especially those containing creatinine and PAH; the other is for 
Withdrawing blood samples. 

A control blood sample is collected, using 2 syringes. The 1st 



0.5 cc of blood drawn is discarded and the sample collected into a 
syringe containing 0.1 cc of heparin solution for each 10 cc of blood 
drawn. The catheter is then flushed with a drop of heparin and 
stoppered. The blood sample is centrifuged at once. If CO 2 and pH 
are to be measured, the syringe should be oiled and should contain 
a drop of Hg, the blood delivered under oil through a long needle 
and centrifuged not more than 5 min. The plasma is then pipetted 
off (directly into Van Slyke-Neill pipets for CO 2 and pH); the red 
cells are suspended in an equal volume of saline, filtered through a 
loose, washed cotton pledget and injected into the animal so that 
red cell depletion does not complicate the experiment. 

The Cr-PAH prime (solution A) is injected, followed by infusion 
of the sustaining mixture (solution B). Thirty min is allowed for 
attainment of sufficient equilibrium to permit accurate clearance 
measurements. A constant rate of sustaining infusion is seldom 
secured by a drip, so a constant infusion pump should be used.§ 

Collect control clearance periods. The bladder should be washed 
thoroughly at the beginning mth 3 or 4 vol of distilled water and 
each tested for completeness of recovery. We collect 3 control 
periods of 20 min each. 

The number and timing of collection of blood samples should be 
arranged to yield the most accurate possible curve of plasma con¬ 
centrations; sampling is done at least once and often twice during 
each urine collection period. The infusion pump is switched off just 
before and during blood collection soas toavoid cross-contammation 
between the catheter orifices; the blood should therefore be with¬ 
drawn as rapidly as possible. , ui 11 ^. 

The saline prime (solution C) may be started during bladdei 

wash-out of the final control period. Volume and rate o* 
tration are predetermined for each experiment- As much as 150 
cc/kg has been injected at a rate of 200 cc/min. Rapid intusions aie 

^conveniently delivered through the 
at slower rates may be given through a Y-tube at the mtus 

“profound changes in renal function an.l urine flow and compj^i- 
+• ra 'lilt from bodv fluid expansion. 1 heir magnitude and \elo 
tTa eli a ed to and'rate of infusion of the saline prime. 

UrinIVoutputs of water and electrolyte increase greatly and com- 

j A Bowman infusion ''".S, hotlw delivery, rate 

Sietw atta“'rv“ls. STmust be Modified to secure a continuous instead 
continuous range of delivery rates. 


were restored by appropriate infusion. As indicated, we have used 2 
methods interchangeably in selecting the infusion, viz., direct 
rapid analysis and estimation of probable composition. For direct 
analysis, the urine is analyzed rapidly for Xa+, and pH as 
soon as the clearance period ends. From pH (assuming a normal 
PCO 2 ), HCO 3 - is estimated. The balance of the anion to match the 
sum of Na+ and K+ is assumed to be the chloride. A replacement 
solution equal in volume to the urine collected may be prepared in 
30 min by this method. The method of estimation of probable 
composition follows from the observation that for any given animal 
and protocol, the urine composition changes according to a definite 
pattern and time sequence. Consequently, “synthetic urines” which 
will not differ greatly from those subsequently obtained may be pre¬ 
pared in advance of the experiment. Their use, volume for volume, 
to replace the succession of urine volumes collected prevents more 
than minor fluctuations in plasma composition. All saline replace¬ 
ments are administered by drip through a Y-tube in the infusion 


Osmotic diuresis, in the broadest sense, may be defined as an 
increase in urine flow caused by increased rate of solute delivery 
into the urine. An increased rate of urinary solute excretion may be 
produced by: (1) Nonionized substances delivered essentially 
quantitatively into the urine as filtered, e.g., mannitol, sucrose. 
(2) Nonionized filtered substances whose rate of delivery into the 
urine is less than that filtered because of losses from the tubule 
through passive back diffusion (urea). (3) Nonionized substances 
filtered in excess of the tubular reabsorptive capacity, e.g., glucose. 
(4) Nonionized substances excreted because of toxic depression of 
normal tubular reabsorption (phlorizhinized glucose). (5) Ionized 
substances delivered almost or entirely quantitatively into the 
urine as filtered, e.g., sulfate, thiosulfate, ferricyanide. (6) Ionized 
substances filtered in excess of tubular reabsorptive capacity, e.g., 
bicarbonate, isotonic or hypertonic saline, xanthine diuresis in part! 
(7) Ionized substances delivered to the urine by tubular secretion, 
e.g., PAH. (8) Ionized substances which cause toxic depression of 
normal tubular reabsorption—mercurial diuresis, xanthine diuresis 

m part. 

Conventionally, the term osmotic diuretic applies to types 1-5. 

* For a review of the literature of this field, see (13). 

is taken by gravity directly into a 
adder catheter at the midpoint of a 



The particular substance whose presence in the urine is primarily 
responsible for the increased flow is termed the osmotic or loading: 
agent. The discussion which follows formulates the general basis 
for a quantitative description of the urine flow during osmotic 
loading by nonionized substances. 

The rate of urine flow is closely correlated with the rate of solute 
excretion where solute is measured in terms of its osmotic activity 
(milliosmols/min). From lowest to highest urine flows, a change in 
solute excretion rate is accompanied by a parallel but not strictly 
linear change in urine flow. Although urine osmolarity may differ 
widely from plasma osmolarity at low rates of solute excretion, the 
variations become less and less as the rate of solute excretion in¬ 
creases. Mathematically, osmolarity of the urine approaches that 
of the plasma as a limiting value as the rate of solute excretion be¬ 
comes indefinitely large. This tendency to approach the plasma 
osmolarity as a limit is observed irrespective of whether the urine 
tends initially to be hyperosmotic or hyposmotic to the plasma. 
Expressed in functional terms, small and limited renal concentrat¬ 
ing and diluting mechanisms become of less and less relative 
importance when they are brought to bear on large and increasing 
quantities of a primary, isosmotic urine, t At lower rates of flow of 
a primary, isosmotic urine, concentrating and diluting mechanisms 
may bulk relatively large, so that large de\'iations from the osmo¬ 
larity of the plasma may be expected. Implicit in this analysis is a 
careful distinction between, on one hand, urine osmolarity which 
approaches equality with the plasma as urinary solute load in¬ 
creases and, on the other hand, activity of concentrating or dilut¬ 
ing mechanisms. Activity of the concentrating mechanism is 
measured as the volume of water reabsorbed/mm without solute 
from an otherwise isosmotic urine. Activity of the diluting mech¬ 
anism is measured as volume of water virtually contributed/min 
without solute to an othermse isosmotic urine. Obviously, if both 
mechanisms should be simultaneously operative, we can measure 
only the difference between them. 

Thus the basis is provided for a separation of the urine A®'' 

L the ^et operation of concentrating and dilating mechan.sma. 
Each portn'^iB capable of separate and. to a '-ge ^ 

pendent variation. The primary isotonic nnne is rlefined a.s the 

t This term is primarily a uidderurfne'wouH hays 

denotes the volume and gpecificallv retain or excrete water. At this 

were no renal Prooe«8e3 operan^^e isosmotic urine as trul.v 





..snuilar Hcarancc, cali-nlatcl as -V.t Tha inox'onients of 


soliile-froo water are measured as the tree water elearaiiee Chjo, (*al- 

ciilated as —_ ^ . V is tlie latc of urine flow' and Posm and 

P osm 

Uosm ai’G the milliosmolar concentration of plasma and urine. The 
free water clearance may have negative values, indicating a net 
concentrating process or positi^'e values indicating a net dilution oi 
the urine. At all times, the urine flow, F, equals the algebraic sum 
of Cosm and CH 2 O. It is not within the scope of this analysis to dis¬ 
cuss the factors wfliich affect CH 2 O, but important among them in 
the normal kidney are the le\'el of neurohypophysial activity and 
the magnitude of Cosm itself. In the presence of maximal antidiuretic 
hormone activity and at low' rates of Cosm, ( —)CH 20 is small and 
is limited by the osmotic ceiling of the urine (in man, 3-4 times the 
osmotic pressure of the plasma). As Cosm increases, ( —)CH20 in¬ 
creases rapidly and appears to approach a limiting value, in man, 
of 5-7 cc/min at an osmolar clearance of 15-25 cc/min (see Brodsky, 
this volume). Further studies are essential to clarify this relation. 

The urinary solute load determines Cosm and therefore the urine 
volume. The solute load during osmotic loading w'ith a nonionized 
substance consists almost entirely of 2 fractions: (1) the loading- 
substance itself, and (2) chlorides, mostly sodium and potassium. 
The rate of delivery of the loading agent into the urine is defined as 
rate of filtration of the agent less the rate of tubular reabsorption 
{U X F = PXCf — T). T \s specific and definable for each sub¬ 
stance. For mannitol, T is small, 10 per cent or less of the filtered 
load, but W'ith the exact limits not clear. For glucose, T is the maxi¬ 
mal rate of tubular reabsorption of glucose {Tnia), term w'hich is 
highly constant in each individual. For urea, T is a mathematical 
function of Pu and Cf/F, increasing as Pu increases, decreasing as 
Cf/V decreases. Since the rate of appearance in the urine of 
nonelectrolytes such as mannitol, glucose or urea is fairly easily 
stated in terms of glomerular filtration rate, plasma concentration 
and tubular reabsorption, the problem of osmolarity as a deter¬ 
minant of urine flow w'ould be readily soh'ed Avere it not for the 
fact that the leading solute is augmented by solute derived from 
im paired reabsorptioii of filtered sodium and chloride.§ No simple 

in (9? andlllT""'''’®" ^letermined by the methods described 

wr‘e itr;Yo'; S°Sa„ge“"proS“l 



reason can be seen why this should be so. The tubular reabsorption 
of other constituents of normal plasma is impaired little or not at 
all. Electrical effects are not involved. There is no correlation with 
composition or molecular structure of the diuretic agent. The in¬ 
creased volume of urine flow consequent on increased solute load 
appears to be accommodated for in part by tubular dilatation and, 
to a much smaller extent, by an increase in mean tubular velocity 
of flow. There is no evidence that either dilatation or increased 
velocity can account for the increased NaCl excretion effected by 
loading with mannitol, etc. Indeed, most conditions which, like 
dilatation, increase intrarenal pressure (edema, venous congestion) 
are associated with increased rather than decreased sodium reab¬ 

By elimination, and on experimental grounds, we may suspect 
that the effect of osmotic load on NaCl reabsorption is an osinotic 
one. It may be surmised that the diluting effect produced within 
the tubule by the retained osmotic agent together with its water 
impedes in some fashion the reabsorption of NaCl and that this 
diluting effect is less important in the case of other plasma con¬ 
stituents. Like the osmotic loading agent, the quantity of NaC 
delivered into the urine depends on the balance^ between filtered 
load and tubular reabsorptive capacity; so that, m effect, osmotic 
loading depresses tubular reabsorptive capacity for JNa and L . 
But the filtered NaCl load contributes equally to excretion so 
that we are not justified in expecting that a given rate of excretion 
of an osmotic agent will necessarily “obhpte the excretion o 
given quantity of NaCl, nor can we quantitatively express, at this 
time the inflLnce of the various factors which determine NaC 
Lcretion during osmotic diuresis. Qualitative expression of some ol 

them^an be supplied ^ leading substance, e.g., urea 

NaCl excSn will decrease if filtered NaCl load decreases and 

VTfi ““t dr 

to be a *Joad^facto^m^^^^^^^^ 

r.T»tubird£ - 


the inhibition of NaCl reabsorption and hence in promoting its 
excretion, it seems reasonable to express this effect in terms of a 
certain limiting plasma/tubular fluid concentration difference and 
to assign this effect to the proximal tubule. 

6. Reasons have been given elsewhere (13) for considering as 
likely the existence of a second NaCl reabsorptive process lying 
distal to the proximal convoluted tubule. This process seems (a) 
to involve much smaller mass transfer than the proximal tubular 
process, but (6) to reabsorb NaCl over a larger concentration 
gradient. Therefore we may suspect that distal reabsorption is not 
materially affected by the presence of an osmotic loading agent. 

7. An osmotic loading agent will deliver increased quantities of 
NaCl into the urine only insofar as, by depressing proximal re¬ 
absorption, it causes the NaCl load to exceed distal reabsorptive 
capacity. Circumstances can be defined in which even a large 
urinary load of an osmotic agent will have little or no effect on 
NaCl excretion. 

8. Variability of the intrinsic or resting tubular NaCl reabsorp¬ 
tive capacity has not entered into the foregoing analysis. Such 
variations do occur in some circumstances. However, it is not 
clear that the variability occurs in the proximal or the distal or in 
both reabsorptive processes, and the occurrence of such variations 
during osmotic loading has not been established. 

Osmotic diuresis produced by loading with ionized substances is 
much more complex than that produced by loading with nonionized 
substances. Sulfate is a good example. If sulfate is injected, all of 
the filtered sulfate, less that very small and fixed amount re¬ 
absorbed by the tubules, appears in the urine. In addition to sulfate, 
the double electrical charge on each ion forcibly extracts into the 
urine 2 cations; sodium provides the largest share of these. Thus, 
in contrast uith nonionized osmotic agents, excretion of sodium 
plus potassium is always either equal to or greater than the equiva¬ 
lent rate of excretion of sulfate, and the minimal rate of total solute 
excretion under sulfate loading is easily predicted. When the sul¬ 
fate load is large, however, chlorides appear in the urine, further 
enhancing the urinary solute load. The mechanism by which this is 
produced is not clear. 

Selection of the appropriate osmotic diuretic for excretion studies 
should be made only after careful consideration of the advantages 
and disadvantages of each. Injection of hypertonic sodium sulfate 
produces an intense diuresis. It also produces Avithdrawal of cellu¬ 
lar water, expansion of the extracellular fluid and decreases in 

Isosmolar solutions of 
this salt, while avoiding cellular dehydration, produce a somewhat 



}>:roater expansion of extraeollular fluid volume. To these ehan}i;es 
are added the diffieulties of describing the mechanism by which sul¬ 
fate impedes chloride, reabsoi’ption. Injection of carbohydrate di¬ 
uretics produces the same general changes in the plasma as the 
saline diuretics except that sodium concentration is lowered in the 
same proportion as Cl" and HCOs". Injection of electrolyte to 
counter these effects results in cellular dehydration and over¬ 
expansion of extracellular fluid. The advantage of urea is that it 
does not as such change body water or electrolyte composition or 
distribution, so that these remain under the investigator’s control. 
Disadvantages of urea are passive back-diffusion of some of the 
filtered load so that diuretic efficacy is diminished, and lack of a 
suitable experimental basis for comparison Avith other diuretics 
because of the extracellular location of most of these and the dis¬ 
tortions of body water and electrolyte wdiich they induce. For 
general studies of the specific effect on electrolyte excretion of 
heavy osmotic loads (more than 30 mOsm/1 of plasma), substances 
other than urea cannot be recommended. 


Our object in this experiment Avas study of renal function and 
factors determining electrolyte excretion under sustained osmotic 
(urea) loads A\dth minimal changes in body fluid volume or com¬ 
position. , 

a) Solutions—A: Cr-PAH prime, as on page 181. 

B: Cr-PAH sustaining solution, as on page 181. 

C; Chloralose;"andsee6) beloAv. 

D • Urea prime. The urea may be made up as a 50 per cent solu- 
tion and filtered. The desired quantity is added to a synthetic 
urine” (solution E-1). This synthetic urine is divided into 2 por¬ 
tions One contains the anesthetic dose of cliloralose, 1 per cent, 
and some of the urea in a concentration not greater than 10 pei 
cent The other contains the. rest of the urea prime m a concen¬ 
tration not greater than 20 per cent. If insufficient fluid is aval able 
for the priming dose of urea, a portion of the 2d synthetic replace- 

'"p- Svnthrtb m'ine^^^ P''"; 

1 q-'ko i<st fF-D is estimated to match volume and 

pared ,v during the injection 

composition oi ^mh tv for each 5 g of 

of P'2T(F-2) matcL the estimated volume and com- 

;':d(ol? of' mil^vMch\,'::^‘he excreted during the 1st 10 mni a,ter 

~ ^ II TTiiHimAii Piiris, l^rs-ncPt ugpnt, II* 

II, pure; ^ York 

P. Ho.ssiKer & C,o., 55 Vaiidam St., Ne'w York. 



injpction of the prime. The 3d (E-3) approximates the mean com¬ 
position of the urine which wnll be excreted during the experiment. 
This solution is added to infused urine to replace those portions 

which will be retained for analysis. 

F: Cr-PAH replacement. 1.0 pei- cent Cr and 0.5 per cent PAH in 

distilled water. 

G; l.OMNaHCOa—10 ml. 

b) Anesthesia.—Anesthesia, is desirable because of the transitory 
distress occasionally produced by the priming dose of urea and be¬ 
cause of the long duration of many of our experiments. Chloralose 
is selected because of its minimal effects on blood pressure, heart 
rate, renal blood flow and filtration rate and the infrequency with 
which it modifies a water diuresis. Since chloralose is not very 
soluble in water, fluids which must be given as part of the experi¬ 
ment should be used as the solvent wherever possible. For the anes¬ 
thetizing dose, the 1st portion of the urea prime may be used. 
Supplementarj'^ doses to sustain anesthesia may use some of the 
fluid replacements. Anesthetizing dose for dogs is 100 mg/kg in a 
1 per cent solution. An additional 20 per cent of this dose should be 
available in a separate container in saline since an occasional ani¬ 
mal requires more than the calculated dose. 

Since chloralose rapidly hydrolyzes to glucose and chloral in hot 
water, it should be dissolved by rapid heating, with stirring, but it 
should not boil. As soon as dissolved, it is filtered rapidly through 
cotton and cooled under tap water to about 40 C. Prepare immedi¬ 
ately before using. 

The 1st half of the anesthetizing dose may be given rapidly, the 
rest more slowly. Ease of induction depends on calmness and re¬ 
laxation of the animal. Chloralose depresses spinal reflexes poorly 
so that accurate estimation of the depth of anesthesia requires ex¬ 
perience. As anesthesia deepens there is progressive loss of co¬ 
ordinated movement and of the more complex reflexes. When anes¬ 
thesia is deep, there remain only isolated, erratic twitches and the 
rnyotatic reflexes. Animals may pass into respiratory paralysis at a 
time when muscle tone is good and the deep tendon reflexes are 
vigorous. We have not yet observed any significant effect of this 
anesthetic on renal function. To maintain a constant depth of 
anesthesia, the animal will need another anesthetizing dose in 
divided portions every 3-4 hr. This may be dissolved in some of the 
synthetic urine replacement (solution E-3) 

c) Procedure-The basic plan of this experiment is similar to 
that described for extracellular fluid expansion (pp 181 ff) I’he 
animal is prepared and 3 control periods are secured. Because of 
tlie lar«e urine flows produced and the more complex urine ,.om- 



position, the animal’s own urine is reinfused to maintain water and 
electrolyte balance. During the control and equilibration periods, 
electrolyte and water losses are negligible and pump infusion of 
solution B maintains plasma concentrations of PAH and creatinine. 
At the end of the last control period administration is begun of the 
anesthetic and some of the urea prime in the 1st portion of solu¬ 
tion E-1, followed by the rest of the urea in the 2d portion of E-1. 
These with solution E-2 replace salt and water losses during urea 
priming and the short (10 min) equilibration period that follows. 

The urine is collected from termination of the last control to the 
end of the equilibration time in 2 time intervals of 15-20 min each. 
Rate of urine flow of the 1st portion is calculated. As the 1st period 
of clearance determination under osmotic load begins, infusion of 
the Cr-PAH sustaining solution is terminated and followed by infu¬ 
sion of the 1st urine sample at its calculated rate of flow. This in 
turn is followed by the 2d urine volume at its estimate rate of flow. 
This sequence of calculation of rate of urine flow and reinfusion of 
the urine at this calculated rate continues through the experiment. 

It wdll be noted that a change from infusion of a Cr-PAH solu¬ 
tion (solution B) to infusion of the animal’s own urine does not 
interrupt the constancy in the minute infusion rate of Cr and PAH 
since their output has come to equal input during the equilibra¬ 
tion and control periods. During replacement of the infusion by 
the urine excreted during the preceding 30 min., input of Cr and 
PAH continues to equal output, except for small, short term, self- 
correcting disparities. Animal, infusion flask and urine receptacle 
are converted into a closed system in which renal function may be 
measured for many hours mth hardly perceptible variations m 
nlasma Cr and PAH concentrations. Our data show no evidence of 
significant accumulation of non-Cr or non-PAH chromogens 
during 8 hr of urine reinfusion. The small amounts of Cr and PA 

lost from the system are restored by solution F. ^ 

As each sample of urine is obtained, the following operations are 
performed as rapidly as possible: calculation of urme mv samphng 
for analysis after mixing, filtration through washed cotton r^ 
placement of estimated loss of urea and electrolyte in 
Lmole by addition of solution E-3 and loss of Cr and PAH by 
solution F* and addition of small amounts of solution G to main¬ 
tain plasma CO, as indicated by analysis of the plasma at intervals 

M"!h“ Tof an"experiment the animal remains under a heavy 
osmot“so that it may become severely dehydrated when urine 

- r rest) X p (est) X ml in urine saigple 

* ml sol. F added *= jQy (ml/min) 


replacement is stopped. To protect against this, a urea-free rnodifi- 
cation of solution E-3 is infused in an amount of 40 cc/g of urea 
used in the priming dose. 

Comment by William A. Brodsky 

It is well known that changes in the volume and composition of extra¬ 
cellular fluids are occasioned by the injection of various loading solutes 
(1). However, the volume of urine during osmotic diuresis is dependent 
only on the number of osmotically active particles reaching the most distal 
portion of the nephron and is independent of (a) changes in volume of ex¬ 
tracellular fluid, (6) changes in pH and CO 2 content of plasma and (c) 
alterations in composition of urinary solutes. The following comparative 
observations will lend support to the aforementioned hypothesis: 

1. Urea, diffusing through all fluid compartments and producing little 
or no change In extracellular volume, evokes a response in urine flow and 
load almost identical to NaCl, a solute distributed through the extracellu¬ 
lar space and producing a significant increase in extracellular volume (11). 

2. Sodium bicarbonate, although provoking profound alkalosis, pro¬ 
duces a diuretic response in dogs closely similar to that of NaH 2 P 04 - 
Na 2 HP 04 , a potent acidifying salt (10). 

3. Sodium sulfate loading, wdth its attendant increase in plasma sodium 
level and suppression of chloride excretion, induces a response in urine flow 
and solute excretion similar to that of mannitol with its associated hypo¬ 
natremia and marked increase of chloride excretion (11). 



6 . 


8 . 

10 . 

11 . 

12 . 






Brodsky, W. A.; Rapoport, S., and West, C. D.: J. Clin. Invest. 29: 1021, 

Folk, B. P.; Zierler, K. L., and Lilierithal, J. L.: Am. J. Physiol. 153: 381, 

Hald, P. M.: in Visscher, M. B. (ed.): Methods in Medical Research 
(Chicago: Year Book Publishers, Inc., 1951), Vol. 4, p. 79. 

Hastings, A. B., et al.:J. Gen. Physiol. 8: 701, 1928. 

Johlin, J. M.: J. Biol. Chem. 91: 551, 1951. 

Ladd, M.: J. Appl. Physiol. 3: 379, 1951. 

Ladd, M., and Schreiner, G. E.: J.A.M.A. 146: 642, 1951. 

Nelson, P. W., and Welt, L. N.: J. Clin. Investigation, in press. 

Peters, J. P., and Van Slyke, D. D.: Quantitative Clinical Chemistry (Bal¬ 
timore: Williams & Wilkins Company, 1932), Vol. II, “Methods.” 

Rapoport, S., and West, C. D.: Am. J. Physiol. 162: 669, 1950. 

Rapoport, S., et al. : Am. J. Physiol. 156: 433, 1949. 

Selkurt, E. E.: in Potter, V. R. (ed.): Methods in Medical Research (Chi¬ 
cago: Year Book Publishers, Inc., 1948), Vol. 1, p. 195. 

Simth K W.: r/ie Kidney: Its Function in Health and Disease (New 
York: Oxford University, 1951). 

Van Slyke, D. D., and Hiller, A.: J. Biol. Chem. 167: 107 1947 

Van Slyke, D. D., et al.:J. Biol. Chem. 183: 331, 1950. 

Wesson, L. G., Jr., and Anslow, W. P., Jr.: Unpublished data. 



IV. Determination of Water and Electrolyte Excretion 
during Osmotic Diuresis in Hydropenic Man 

WILLIAM A. BRODSKY, University of Louisville 

The responses of the kidney to solute loading during the hydro¬ 
penic state provide an excellent means of measuring renal capacity 
for water and electrolyte conservation. During osmotic diuresis in 
man (12) and dog (15), a constant relation between urinary flow 
and excretion of solutes is observed with a wide variety of loading 
solutes. The pattern of electrolyte excretion is reproducible for 
any single loading substance. Rapoport and West (7) have shown 
that imposition of appropriate loads of the sodium salts of CNS, 
NO 3 , HCO 3 , PAH, SO 4 , PO 4 , Fe(CN )6 and S 2 O 3 in each case alters 
the pattern of chloride excretion in a specific manner, dependent on 
interionic antagonism and predictable from the Hofmeister series. 
On the other hand, loading with nonelectrolytes (mannitol, xylose, 
sorbose, glucose, sucrose) results in a pattern of electrolyte loss 
independent of interionic antagonism and representing a stress¬ 
load curve of electrolyte conservation during the hydropenic state. 
To standardize conditions for the assessment of water and elec¬ 
trolyte economy during osmotic diuresis, a nonelectrolyte such as 
mannitol is used as the loading solute. Mannitol has the advan¬ 
tages of innocuousness, availability and ease of deteimination. 


a) Preparation of subject .—Water and food are withheld foi 
12-16 hr before the test so that urine flow is minimal and osmotic 
activity maximal immediately before injection of the loading solute. 
A priming injection of mannitol, in 25 per cent solution, is given 
intravenously in an amount calculated to add 20M0 mOsm/1 of 
distribution volume, followed by a maintenance ^fusion m t e 
amount of 1 per cent of the priming dose/mmMf simultaneous 
measurement of renal plasma flow is desired, PAH is added to the 
infusion fluids in quantities sufficient to maintain a plasma ^ve 
3 mg /lOO cc. (The amount of PAH used to measure effective 
renaf plasma flow is too small [concentration of 3 mg/100 cc is 
u 4 - f 1 mOsm/11 to interfere with electrolyte excietion 

,oa,.. 120-40 n.Os./ll of .aonito., 


Sample Protocol for Mannitol Loading 

Patient: John Doe, age 25 years weight 70 kg 
Level of mannitol desired: 20 mOsm/1 



Assumed vol of distribution: 0.2 X 70 k}^ body weight 

Amt of mannitol required: 20 X 14 = 280 mOsm or 51 g; or 204 cc of 25% 


Disappearance rate: 1% of amt in the body/min 
Corrected priming dose: 214 cc 

Maintenance dose: 2.04 cc/min . , n j 

Correction for amt of mannitol e.xcreted during a 10 mm period allowed 
for priming injection; 204 (0.5) X (0.01) (10) = 10.2 cc 

h) Constant infusion machine.—For this procedure, the use of a 
constant infusion pump is most desirable. The apparatus used in 
our laboratory employs 2 electrically regulated pistons which push 
the plungers of 2 calibrated syringes, the smaller of which is used 
for slow infusion of potent pharmacologic agents.* 

c) Collection of urine and blood samples. —Urine is collected by 
catheterization in females and by voluntary voiding in males. 
Before loading, urine is collected over 20-30 min intervals for 2 
periods. After the 10 min priming injection, 15 min is allowed for 
equilibration of body fluids toward a steady state. The equilibra¬ 
tion urine, designated X — 1, represents the time interval from 
the last preliminary period to the end of equilibration. After 
equilibration, with urine flows at diuretic levels, the collection 
periods are of 10-20 min duration. Three or 4 such samples are col¬ 
lected. Blood samples are drawn at approximate midpoints of col¬ 
lection periods. 

The entire procedure is readily tolerated by the patients and lasts 
no longer than 2 hr. 

d) Definition of terms. —1. Osmotic Activity = concentration of 
the total number of osmotically active constituents in urine cal¬ 
culated by the degree of depression of freezing point or vapor ten¬ 
sion (Raoult’s law). 

2. Effectiv'e Osmotic Load = the total number of osmotically 
active particles excreted each minute; the product of the minute 
\'olume and osmotic activity of urine. 

3. Flow-Load Relationship — the relation between minute vol¬ 
ume of urine and effective osmotic load. For example, in the hydro- 
penic state, a constant volume of urinary water is required for 
excretion of a given osmotic load. 

4. Particulate Concentration = the total number of particles 
per unit volume of solution as determined by the sum of observed 
concentration of each solute. Particulate concentration will differ 
fiom^osmotic activity (especially in electrolyte-containing solu- 

eialfe Comia.'^yl 13 Ave"! ^pe- 

pump. obtainable from Process and Instruments. 60 GreenpoinriS^Troildy™ 



tions) to an extent predictable by the degree of dissociation of the 
salts and forces of interionic attraction. The term is synonymous 
with “true total osmolarity,” or the osmolarity calculated at infi¬ 
nite dilutions. 

e) Determination of osmotic activity. —According to Raoult’s 
law, the freezing point depression of a dilute aqueous system is 
proportional to the mole fraction of solute in solution and there¬ 
fore to the osmotic activity of the solution. The principle and 
technique of measuring freezing point depression of solution have 
undergone little change since first used by Beckman in 1888. 

URINE. —The apparatus for determination of freezing point depres¬ 
sion, commonly used, is the Cryoscope.* It consists of a large test 
tube of 100 ml capacity placed in a steel jacket surrounded by a 
Dewar flask. An aliquot of urine is placed into the test tube fitted 
with a stirrer and a Beckman thermometer graduated in hundredths 
of a degree (range, 4-2 to —2° C). The Dewar flask is Vs to V 2 
filled with ether. Water suction applied to the ether-filled jacket 
vaporizes the ether, thus cooling the solution in the inner tube. 
By gradual lowering of the temperature of the liquid and stirring 
to avoid excessive supercooling, the freezing point, at which the 
temperature remains constant, can be observed. Osmotic activity, 
expressed as mOsm/1, is obtained by applying the formula 

. . freezing point depression X 1,000 
Osmotic activity = --- 

where 1.86 is the molal depression constant for aqueous solutions. 

PLASMA. —If sufficient plasma is ax'ailable, direct determination 
of the freezing point is possible. This technitiue, however, has not 
in my experience been highly accurate in protein-containing solu¬ 
tions. Consequently, osmotic activity of plasma is estimated from 
the sum of the determined solutes from the equation 

Osmotic activity = 2(Na -[- K) - 8 -f- (urea) -h (loading solute) X 7 

where 7 = 0.925, the assumed activity coefficient of plasma, and 
the terms in parentheses are in the following units: Na and K, in 
mEq/1, so that 2(Na -f- K) - 8 = the osmolar concentration of 
electrolytes; urea, loading solute and osmotic activity, in mOsm/ 
f) Measurement of osmotic limitations of the kidney. Iipiie , « 
sample plot of flow vs. load, illustrates the dependence of urinary 
flow on the effective osmotic load during^ osmotic diuiesis vit 
mannitol in man. Since similar patterns of flow 
served with many different loading solutes, it 
that water reabsorption must be the final event of tubular fu 

* Manufactured by Eimer and Amend, New York, N. Y, 


tioii during hydropenic diuresis (1, 12, 15). Thus the flow-load 
cur^’■e may be used to delimit distal tubular function. Data falling 
to the left of the curve indicate a defective capacity for water con- 

of a gh°en ILd"* 

In the production of hypertonic urine the osmolniM+’ir rvf u* \ 
exceeds that of body fluids, a water economy fH n t f ' 
It may be expressed as (2, 7) ^ ( 2 ec) must result. 

Effec tive osmotic load 

^smotiU^^lty^n)lasma ~ = IEOec 



The value of the 1st member signifies the voluma of isotonic body 
fluid cleared by the kidney. It may be interpreted also as the 
vmlume of isotonic fluid presented to the distal tubule after proxi¬ 
mal iso-osmotic reabsorption. The water economized distally 
during osmotic diuresis is distributed among the body fluid com¬ 
partments so that they tend to return toward the preloading level 
of osmotic activity. Figure 2 illustrates the relation of water econ¬ 
omy to urine flow during mannitol diuresis in man. A level of 5-7 
cc/min is reached at flows exceeding 4 cc/min, with small change 
at higher flows. The intensity of water economy, i.e., the amount 

conserved/cc. of urine elaborated is greater, the higher the osmolal¬ 
ity of the urine. Thus, at extremely high flow, when the osmolarity 
of urine approaches that of water, intensity of water economy will 
approach very low levels. 

q) Measurement of electrolyte excretion—1. assessment oi 

chloride loss during osmotic diuresis in man is almost linear y le- 
lated to urinary flow or load (2, 9). Since total particulate and ef¬ 
fective osmotic loads are almost identical when the predominant 
urinary solute is a nonelectrolyte, the plot of electrolyte losses vs 
effective load willMepict graphically the 

solutes over a wide range ot excreted loads. lor examp e, Figu 
illustrates the constm.t patter.,s of sodium, a,.d 
loss/unit of excreted solute in 1() normal subjects and in d patien . 
with diabetes insipidus under similar loading conditions. 1 he sum- 
llrUy S the "a«er„ h. diabetes iusipidus to the no.mal ...d.cates 


that solute excretion proceeds in a perfectly normal mariner de¬ 
spite the gross defect in water excretion (1). 

From data on solute composition and flow-load relationship in 
the urine, one may localize approximately the site of renal defects 
in various clinical states. For example, in diabetes insipidus, the 
finding of an abnormal flow-load relation with a normal composi¬ 
tion of urinary solutes suggests a distal portion of the tubule as 

</) 2 

O ^ 

_l N. 

2 ” 
(A E 

<0 V 
2 ^ 

0 20 


3 4 5 

LOAD-mOsm/min /173 






• . *0 0 


t ' I '->---1-'-T-'-r 

« T --' -*" 

2 3 4 5 

LOAD mOsm./mia/|.7 3M^ 


LOAD mOsm./min./l73Mi 

sodium losses vs. osmotic load during 
mannitol loading. Open and solid circles refer to 3 patients with diahptp« In 

X: X r* ™-Po«ition of Ly 

solute.s suggests proximal tubular defects. It is of interest thit 

leithei the pattern of electrolyte loss nor flow lo-nl if^lMi i 
..e™ .-on-elafe.! with the le.el Jif 


Another approach to the study of electrolyte exeret 10^00"IsistX 



observations on concentration gradients from plasma to urine 
reached by given constituents. For example, during infusions of 
hypertonic saline the concentration of urinary chloride reaches 
levels as high as 300 mEq/1, whereas during water diuresis the 
concentration falls to 5-10 mEq/1. In certain clinical states of 
“salt-losing,” loading with hypertonic saline fails to increase the 
concentration of urinary chloride above 150 mEq/1, whereas water 
loading fails to depress urinary chlorides to low levels (10). 

3. WATER DIURESIS. —Water diuresis is a condition during which 
electrolyte conservation proceeds at maximal levels (4, 5, 6, 14). 
Studies of water and solute excretion under such conditions should 
yield data describing capacities for both electrolyte conservation 
and dilution functions of the kidney. 

4. BALANCE STUDIES. —These may be used as a measure of renal 
capacity for electrolyte conservation in the presence of hyper- or 
hypoelectrolytemia, with controlled intakes of salt (10, 11, 13). 

5. THERMODYNAMIC WORK. —Calculations of the “ideal” or 
thermodynamic energy output of the kidney, under conditions of 
large osmotic loads, may serve as a useful parameter of the renal 
capacity for solute excretion. Such calculations have limited value 
because osmotic energy is but a small fraction of the total energy 
output of the kidney; nevertheless the pattern of osmotic work dur¬ 
ing solute loading is constant, approaching a biologically maximal 
value (2 8). Calculations of work necessitate addition of the mini¬ 
mal free energy necessary to effectuate the observed concentration 
gradients of all solute constituents from plasma to urine. Applying 
the formula of von Rhorer (13) 

W = VRT (Uln U/P P - U) 

where R is the gas constant; T the absolute temperature; W the 
energy expenditure in g-cal/min; V the corrected minute volume 
of urine, and U and P the molar concentrations of each 
urine and plasma, respectively. It has been found f a* osmobc work 
in hvdropenic man is of the order of approximately 0.4 g-eal/mm/ 
i 73 mU 8 ) and that it rises to a level of 3-4 g-cal after loading (9) 
Further studies are needed to assess the value of such calculatioi 

in various disease states. 

Note.— This section was reviewed by Robert F. Pitts. 

Comment by Clark D. West , • i, ^ if 

^h^ul'^eXx" mhable indicator of presence of liydropenia is a 


low rate of urine flow in the preliminary periods (0.5 cc/min or less) and a 
high osmolarity (greater than 1,000 mOsm/1). 

Observations on urinary composition and volume after solute loading, 
first made by McCance and later and more extensively by Rapoport, have 
led to new and interesting concepts of the process of urine formation. As 
Dr. Brodsky indicates, these studies have made clear a functional division 
of the tubule as concerns water and solute reabsorption. Later work, ap¬ 
plying the technique to patients with diabetes insipidus, has raised im¬ 
portant questions as to the renal mechanism for the production of a dilute 

The possibilities for further use of this method of study are numerous. 
The flow-load relation during osmotic diuresis would seem well adapted to 
the study of the action of diuretics. Comparatively little work has been 
done to investigate the flow-load pattern in the numerous clinical condi¬ 
tions in which renal function is abnormal. Perhaps even more valuable 
will be the application of solute loading to study of the multiple factors 
which influence electrolyte excretion. The stable and reproducible electro¬ 
lyte composition of the urine during osmotic diuresis, difficult to achieve 
by other methods, has already been fruitfully employed in study of the ef¬ 
fect of splanchnicotomy on renal electrolyte control and to some extent 
in study of the electrolyte retention accompanying edema. The procedure 
is innocuous especially when nonelectrolyte loading solutes are used, and, 
if elevation of plasma osmolarity is not greater than 20-30 mOsm/1, pro¬ 
duces no more discomfort than the more commonly used tests of renal 
function. Its clinical applications have been too meager to permit comment 
on its value as more than a research tool. 

Comment by Laurence G. Wesson, Jr. 

The procedure might perhaps be simplified for clinical purposes if it 
could be shown that subcutaneous injection of Pitressin (say, 500 mU) 
could substitute for the natural hydropenic state. 

The postulate that “water reabsorption is the final event of tubular func¬ 
tion during hydropenic diuresis” implies that water reabsorption is per¬ 
formed distally to acidification and ammoniation. Can this be supported? 

In the discussion of water economy, the terms “volume of isotonic body 
uid cleared by the kidney and “water economy” correspond respectively 
to our terms^^ osmolar clearance” and “negative free water clear^ce.” 

The term pattern of electrolyte loss” is not self-explanatory. If by it 
IS meant the ratio of rates of electrolyte excretion to eLetion of loaL! 
substances, we have found definite alterations in this term in the dog in 
which glomerular filtration rate is more readily varied than in man. 

Last, can measurements of urinary specific gravitv siibcsfifnfo 
copy in clinical assessment of flow-load funetioJ for cryos- 

Comment by Dr. Brodsky 

That water reabsorption occum^’Sly to toSbnf 



cated by the constancy of flow-load relationships under loading with l)()th 
alkalinizing and acidifying salts. 

The “pattern of electrolyte loss” is the rate of excretion of a specific 
electrolyte in relation to total solute excretion. 

I doubt that measurements of specific gravity can be substituted for 
those of freezing point depression, since the term desired is the relationship 
between urine volume and moles rather than grams per unit volume. 

1 . 

2 . 



6 . 


8 . 


10 . 

11 . 

12 . 



15 . 


Brodsky, W. A., and Rapoport, S.; The mechanism of polyuria of dia¬ 
betes insipidus in man: The effect of osmotic loading, J. Clin. Invest. 
30: 282, March, 1951. 

Brodsky, W. A.; Rapoport, S., and West, C. D.: The mechanism of gly- 
cosuric diuresis in diabetic man, J. Clin. Invest. 29: 1021, August, 

Kaplan, S. A., and Rapoport, S.; Urinary excretion of sodium and chlo¬ 
ride after splanchnicotomy: An effect on the proximal tubule. Am. J. 
Physiol. 164: 175, 1951. 

Klisiecki, A., et al.: The absorption and excretion of water by the mam¬ 
mal: I. Relation between absorption of water and its excretion by the 
innervated and denervated kidney, Proc. Roy. Soc., London, s.B 
112: 496, 1933. 

Klisiecki, A., et al .; The absorption and excretion of water by the mammal: 
II. Factors influencing the response of the kidney, Proc. Roy. Soc., 
London, s.B 112: 521, 1933. 

Marshall, E. K., Jr.: The influence of diuresis on elimination of 
urea, creatinine and chloride, J. Pharmacol. & Exper. Therapy 16: 141, 

Rapoport, S., and West, C. D.: Ionic antagonism: Effects of various an¬ 
ions on chloride excretion during osmotic diuresis in the dog. Am. J. 
Physiol. 162:668, September, 1950. ^ , 

Rapoport, S.; Brodsky, W. A., and West, C. D.: Excretion of solutes and 
osmotic work of the “resting” kidney in hydropenic man, Am. J. Phys¬ 

iol. 157:357, 1949. ^ ^ _ , 

Rapoport, S.; West, C. D., and Brodsky, W. A.: Excretion of solutes and 

osmotic work during osmotic diuresis of hydropenic man. The idea 
and the proximal and distal tubular work; the biological maximum of 
work Am. J. Physiol. 157: 363, 1949. 

Rapoport, S.; West, C. D., and Brodsky, W. A '®“'Oo8mK oond'tion^^ 
The renal defect in tuberculous meningitis, J. Lab. & Clin. Med. 37. 

Rawo^r's-.^e^ab Hypoelectrolytemia in peritonitis, J. Ciin. Invest. 

RapopXs.!'!'! al.: Urinary flow and 

diuresis in hydropenic man. Am. J. ? hyjioh . 56-«3 “arclu _ 
von Rhorer, L.; Ueber die osmotische Arbeit der Nieren, Arch, ge . 

Vern4“ e' B : AbsorpZ'and excretion of w.ater: Antidiuretic hormone, 

w'rc‘D Ind Rlopor^Sd Urine flow .and .solute excretion of hydro- 

^enic dJ^i’under “resting” conditions and during ... Am. 

J. Physiol. 163: 159, October, 1950. 


V. Bioassay of Antidiuretic and Diuretic Substances 


J. MAXWELL LITTLI'L Bowman Gray School of Medicine 

During development of a bioassay procedure for either an anti¬ 
diuretic or a diuretic substance the principal difficulty encoun¬ 
tered has been the insuring of a relatively constant state of hydra¬ 
tion in the test animals, as well as a reproducible response to a fluid 
load presented at or about the time of administration of the mate¬ 
rial being assayed. Several methods of hydration and analysis 
of the response have been published for the white mouse, Avhite 
rat and dog. 

The rat technique for bioassay of the activity of preparations of 
the posterior pituitary was introduced in a quantitative procedure 
by Burn (I). Since then, many modifications have appeared. 
Ham (5) studied several of these modifications (3, 8, 12, 13) and 
found that the method of hydration used by Gilman and Goodman 
(3) and the method used by Krieger and Kilvington (8) for express¬ 
ing the results of the assay gave the most reproducible figures in 
control experiments. Ham and Landis (6) have incorporated these 
principles in the bioassay procedure given here. 


A stock of male albino rats is kept in small groups in a room 
maintained at an even temperature. The animals are fed a weighed 
amount of food in excess of the expected daily requirements. The 
daily consumption is determined by weighing the amount not con¬ 
sumed. Animals not consuming a normal amount of food are not 
included in the assay groups. Fresh water is kept constantly a\mil- 


Before each assay 18 rats are fasted for 12 hr but are allowed to 
have water ad hb. The next morning the rats are nla.cpd bv 

F^ch rat is given by gavage a priming dose of 0.2 

pel- cent NaCl 



solution equivalent to 2.5 per cent of the body weight. Two hr 
after the priming dose was given, all urine excreted up to that time 
is discarded. Each rat is then given a hydrating dose of 0.2 per cent 
NaCl solution equivalent to 5 per cent of the original body weight. 
Simultaneously the solution to be assayed is injected intraperi- 
toneally in a volume totaling 1 cc/100 g of rat. The controls are 
given similar gavage and they receive intraperitoneally either 0.9 
per cent NaCl solution or distilled water to match the composition 

of the solutions being tested. 

Starting then at zero time for each cage, the cumulative volume 
of urine excreted by each group is recorded at 15 min intervals for 
3 hr. The recorded volumes are computed in terms of the amount 
excreted per 100 g of rat and plotted against time for each cage 
on a chart in which 1 in. represents 30 min and 1.0 cc of urine. 

To express the results by a single number, the areas of these curves 
are measured in square inches, using a planimeter. 

Critique .—In 130 experiments the variation between 2 control 
cages, observed on the same day, averaged 1.56 sq in., with a maxi¬ 
mal variation of 5 sq in., which occurred in only 2 experiments (6). 
The method is accurate for doses of 0.5-10 mU of Pituitrm/lOO g 
of rat. As long as the rats eat well and maintain their weight they 
may be used for not more than 3 assays at intervals of 10 days or 

It has been suggested (5, 13) that Pituitrm solutions may be as¬ 
sayed more accurately by measuring the effect of the Preparation on 
the total chloride excretion, expressed as microequivalents/lOO g 
of rat/3 hr, than by measuring the effect on water excretion. With 
the chloride excretion procedure it was shown 
rate range for Pituitrin solutions was from 0.1 to 100 mU/lOO g ot 
rat However, since there is some doubt whether or not the a t - 
diuretic and chloruretic effects of pituitary extracts may be at¬ 
tributed to the same molecular substance (4,10,11) and since it has 
been reported (6) that an antidiuretic substance found m human 
urine wL withmit chloruretic effect, it would appear reasonable to 
restrict the assay procedure to the renal function being studied, 
e g , water reabsorption or chloruresis, as the case may be. 

assat using dogs . 

The principle involved in the procedure of Hare el al (7) 

1 / fnn+mn of changes in the creatinine urinary-plasma (U/P) 
SruTng tm L^ction of an antidiuretic substance .n dogs 

with diabetes insipidus. i female dogs by surgical 

Permanent polyuna is Tained to lie 

transection of the pituitary stalk. The dogs aie 



quietly on padded boards with loose restraining thongs. Food is 
removed 24 hr and water 3 hr before the experiment. Three g of 
creatinine in 100 cc of water is administered by stomach tube 2 hr 
before the assay. Urine is collected by a retention catheter (no. 16 
Pezzer), which is left in the bladder throughout the experiment, 
draining into a volumetric flask. The bladder is emptied at the end 
of each clearance period by washing with air. Blood samples are 
collected at the midpoint of the 1st clearance period and hourly 
thereafter. Creatinine is determined on tungstic acid filtrates ol 
plasma and diluted aliquots of urine by the alkaline picrate method 
(2), and the concentration of creatinine in the urine is compared 
with that of plasma at the midpoint of the clearance period, ob¬ 
tained by interpolation on a plasma concentration-time curve con¬ 
structed from the data on the hourly blood samples. 

A control period of 20 min is run, the material being assayed is 
given intravenously and two 20 min assay periods are run. The 
assay of an unknown should be preceded and followed by the same 
dose of a standard antidiuretic preparation (Pituitrin or Pitressin). 
The usual assay requires 3 hr and 20 min. 

The results are calculated as percentile change in the creatinine 
U/P ratio, taking as the basal creatinine U/P ratio the average of 
the ratios for the control periods immediately preceding and fol¬ 
lowing the assay of the unknown or the standard. Presence or ab¬ 
sence of antidiuretic material may be detected by use of a single 
assay; however, assays are usually run in duplicate or triplicate. 
When duplicate assays agree within 0.3 mU, the average is con¬ 
sidered acceptable; when 2 of the triplicate assays agree within 0.3 
mU and the 3d is erratic, the average of the 2 is considered accept¬ 

The useful range of dosage for a standard is 0.3-3.0 mU. Doses 
of 0.1 mU can be detected, but they may be estimated somewhere 
between 0.05 and 0.2 mU, and doses of 3.0 mU or more never assay 
greater than 3.0 mU. If an unknown gives a value larger than 2.0 
mU the assay should be repeated with a smaller dose. 

Only fleshly collected urine specimens should be used and even 
when they are clear the dogs may retch. These reactions do not af¬ 
fect the assay. Urine from patients or animals with urinary tract 
infections may cause prolonged oliguria or anuria. An occasional 
urine may show a decided diuretic effect (see also (9)) w^hich mav 
prohibit the evaluation of subsequent injections. When the urine 
flow of the assay dog is less than 1.5 cc/min, the maximal dose of 

decreases as the urine flow 

1 1; ’ that the basal urine flow be at least 

1.5 cc/mm, and one of 2-5 cc/min is preferable. 



1. Hum, .1. II.: C^uart. .1. Fharm. Pharmaca)!. 4: 517, 

2. Foliu, O., and Wu, H.: .1. Biol. Chem. 38: 81, 1919. 

3. Gilman, A., and Goodman, A.: J. Pliypiol. 90: 113, 1937. 

4. Grollmaii, A., and Woods, B.: Endoc.rinology 44; 409, 1949. 

5. Ham, G. C.; Proc. Soc. Exper. Biol. <fe Med. 53: 210, 1943. 

6 . Ham, G. C., and Landis, E. M.: J. Clin. Invest. 21: 455, 1940. 

7. Hare, A., et al.: Endocrinology 36; 323, 1945. 

8 . Krieger, V. I., and Kilvington, T. B.; M. J. Australia 1: 575, 1940. 

9. Little, J. M.: J. Pharmacol. & Exper. Therapy 91; 124, 1947. 

10. Little, J. M., et al.: Am. J. Physiol. 151; 174, 1947. 

11 . Ralli, E. P., et al.: Am. J. Physiol. 163: 141, 1950. 

12. Robinson, F. H., and Farr, L. E.; Ann. Int. Med. 14; 42, 1940. 

13. Silvette, IL: Am. J. Physiol. 128: 747, 1940. 


LOUIS B. TURNER arul ARTHUR GROLLMAN, Southwestern University 

The role of the posterior pituitary hormone in regulation of 
water excretion by the kidney is well established. Because ot this 
relationship, investigators of renal function are often interested in 
determining the antidiuretic action of the urine in order to evaluate 
the role of the posterior pituitary in modifying water exchange. 
It must be emphasized that the substance in the blood and urine 
which exerts antidiuretic action has not been demonstrated to be 
identical with the hormone derived from the neural lobe of the 
hypophysis. In fact, the urinary principle differs in many respects 
from the hormone, lacking, for example, the chloruretic action ol 
the latter. This has been interpreted by Grollman and Woods (4). 
as being due to oxidative processes in the body which abolish the 
chloruretic without affecting the antidiuretic action of the hormone. 
However, as demonstrated by Croxatto, Rojas and Barnah (2), 
digestion of globulin with pepsin results in the production of a po- 
tent antidiuretio compound. It is quite po.ssible, therefore, that 
much of tlie autidiuretic sub-stauce encountered m tlie urine or 
other body fluids is of extrapituitary oiigin. 

The available methods consist essentially in purification ot 11 
urine to a degree whicli permits its injection into liydrated rats and 
determining the antidiuretic action thus induced Earlier prooe- 
ures involved the concentrating of the urine and dialysis ot the 

concentrate. The latter process, f--ytt^ihtvTf tte 

able degree of antidiuretio activity due to ^ 

nrinciole (3). To avoid this source of error and at the sam . 
effect a sufficient purification and concentration ol the ac n e ac o , 
CnSfimae and Wmids (4) have utilised the following procalnre. 



The sample of freshly collected uriiie is filtered and brought t<> 
pH 4 . 5 - 5.0 by the addition of acetic acid. One g of absorbent pow¬ 
dered charcoal (Darco actix'ated carbon, grade G-60, or other grades 
of charcoal suitable for adsorption from aqueous solutions) is added 
for each 100 cc of urine and the mixture agitated at intervals for 
several hours. It is placed in the ice-chest overnight and the super¬ 
natant urine discarded. The charcoal is collected on a Buchner 
funnel, washed with small amounts of distilled water to remove ad¬ 
herent urine, drained and transferred to a cylindrical centrifuge 
tube. The charcoal on which the antidiuretic hormone is quantita¬ 
tively adsorbed is suspended in glacial acetic acid (5 cc/g of char¬ 
coal), agitated Avell and precipitated by the addition of 10 \’ol of a 
mixture of equal parts of absolute ethyl alcohol and petroleum 
ether. The amount of precipitate formed at this step of the proce¬ 
dure is so minute that it is scarcely perceptible, so no attempt shoukl 
be made to remove it by filtration. However, after standing in the 
ice-chest overnight it deposits on the walls of the centrifuge tube, 
and any undeposited material is removed by centrifugation, the 
supernatant fluid being discarded. After drying in a ^’■acuum desic¬ 
cator at room temperature for at least 1 hr to remo\’'e the organic 
solvents, the precipitate is dissolved in a small amount of distilled 
water and filtered. 

It is then assayed by the bioassay procedure of Burn ( 1 ) as 
modified by Ralli et al. (5). Four adult male rats, weighing 120 - 
240 g, fasted 12-18 hr but allowed water ad lib up to the time of 
the assay, are placed in a metabolism cage. To each rat is ad¬ 
ministered 5 per cent of its body weight of lukewarm tap water by 
stomach tube. Immediately, thereafter, 1 cc of the solution to be 
assayed is injected intraperitoneally. The volume of urine ex¬ 
creted is recorded at 15 min intervals, and the time required for 
excreting 50 per cent of the administered water is used as an index 
of the antidiuretic potency of the injected extract. 

Note.— This section was reviewed by J. :Maxwell Little. 


1. Bum, J. H.: Rstimation of aiitidiuretic potency of pituitary (])osterior 

^ lobe) extiact. Quart. .1. Pharm. & Pharmacol. 4: 517 1931 

2. Croxatto, H.; Rojas, G and Barnafi, L.t An antidiu’retic substance ob- 

1951 ' globulin with pepsin. Science 113: 494, Apr. 27, 

3. Elhs, M E., and Grollman, A.: The antidiuretic activity of the urine in 

chmcal and e.xpenmental hypertension. Endocrinology, 44: 415, May, 


4. Grollman, A., and Woods, B.; A new procedure for the determination of 

the antidiuretic principle in the urine. Endocrinology 44; 409, May, 

5. Ralli, E. P., et al.: Factors influencing ascites in patients with cirrhosis of 

the liver, J. Clin. Invest. 24; 316, May, 1945. 


J. MAXWELL LITTLE, Bowman Gray School of Medicine 

Male albino rats weighing 140-240 g are fasted and deprived of 
water for 18 hr before the assay. The animals are divided into 
subgroups of 4 each and placed in small cylindrical metabolimeter 
cages, 10 in. in diameter and 11 in. high, which rest on suitable 
glass funnels. The urine is collected by drainage into graduate 
cylinders. A small wire screen inserted in the neck of the funnel 
prevents feces from passing into the cylinder. Two subgroups 
(8 rats) are used for each dose of a standard or unknown substance. 

Preliminary experiments with an unknown substance are done 
on 1 day, using 5 groups of 8 rats each of approximately the same 
weight. The control group is given 25 cc/kg of 0.9 per cent NaCl 
by stomach tube, containing no drug, and the other 4 groups are 
given geometrically increasing doses of the substance being tested. 
Thus the right dose range is determined in very few experiments. 
If the drug is to be administered orally it is dissolved or suspended 
in 0.9 per cent NaCl and given in a dose of 25 cc/kg. If the ding is 
to be administered parenterally the same saline dose is given an 
the drug is injected immediately afterward. The rats remain m the 
metabolimeter cages mthout food or «-ater for the duration of the 
assay. With many substances the assay may be teiminated at t 
end of 5 hr, since the rate of urine flow will show a sharp dechn 

by this time; but with some substances, such as 

it is suggested that the assay time be extended to 24 hi. On ter 

mination of the assay, residual bladder urine is expelled by pulling 

at the base of the rat’s tail. , j 

For the estimation of potency, 5 groups of 8 rats are us^d « 

day One group receives saline alone, 2 groups receive 2 suitable 
day. une gio p which is used as a standard 

"i^sTv din h^^sll a 2 groups rece^. 

rsultabk Ises of the unknown o/d 

actions will fall within the same ,uflieiently ac- 

ruteT«"of'lt mel^^ ^he substance 

relative to urea. 


The volume of urine excreted during the assay time is expressed 
as per cent of the liquid administered, and it is called the diuretic 
effect. The best-fitting straight line for each set of data is calculated 
from the regression equation 

y = y + h{x - x) 

where y is the mean value of y (log effect), x the mean value of x 
(log dose) and h the regression coefficient determined from the 

_ Sy(x - x) 

S(x — xY 

When the dose-effect lines are parallel, within the experiment 
error, the distance along the abscissa between the line for urea and 
that for the unknown substance gives the logarithm of the diuretic 
activity of the substance relative to urea, with an assigned potency 
of I.O. When the dose-effect lines differ significantly in slope the 
relative potency is computed at a specified level of log per cent 
excretion. With this assay procedure some diuretic potencies, 
relative to urea I.O, were: Salyrgan, 400; theophylline, 115; caffeine, 
32, and ammonium chloride, 2.7 (1). 

Modification .—In our laboratory (2) a modification of this pro¬ 
cedure has decreased considerably the variability of the control 
per cent excretion rates. Food is removed 18 hr before the assay, 
and at that time the rats are given a hydrating dose of 25 cc/kg of 
0.9 per cent NaCl. Water is left available until 1.5 hr before the 
beginning of the assay. From this point the procedure we have 
used is identical with that of Lipschitz et al. (1). Comparisons of 
data obtained by us using this modification wuth those obtained 
using the Lipschitz procedure showed the mean coefficients of 
variation for 6 groups studied on 6 separate days at several time 
intervals to be, respectively: 5 hr, 16.4 and 21.9 per cent; 10 hr, 
11.0 and 18.1 per cent; 24 hr, 12.0 and 14.4 per cent. We have 
also found it advantageous to keep our animals in a room with 
temperature control during the assay period. 


I emale dogs Aveighmg 6-12 kg are prepared for easy catheteri¬ 
zation by an episiotomy, and they are trained to lie quietly on 
the animal board with loose restraining thongs. At 4:00 p.m. on 
the day before the assay is to be carried out they are given 25 cc/kg 
of tap water by stomach tube as a hydrating dose, and food is 

TtTn' nn available until 8 :30 a. m. on the day of assay. 

At 10:00 A.M. the animals are catheterized, the bladder is emptied 



and the urine is discarded, the catheter being left in place. At the 
same time a tap water load of 25 cc/kg is given by stomach tube. 
At 11:00 A.M. the 1st collection period is ended and the unknown 
substance administered. At noon the 2d collection period is ended 
and the animals are returned to standard metabolimeter cages 
without food or water. At 4:00 p.m. the bladder is emptied by 
catheterization, ending collection period 3. The total urinary ex¬ 
cretion for the 3 periods is used to calculate the excretion in terms 
of cc/kg for 6 hr. The duration of assay may be prolonged to 
accommodate a long-acting diuretic substance. The animals are kept 
in a room maintained at approximately 22 C during the assay. 

The control rate of excretion, giving 2 ml of distilled water intra¬ 
venously instead of a diuretic substance, is determined in 8 or 
more experiments for each animal. We have found it convenient 
to use a group of 12 dogs, 6 of which are used on alternate days for 
assay. Sixteen control experiments each on 10 dogs gave the follow¬ 
ing mean urine flows in cc/kg, Avith the corresponding standard de¬ 
viations, for the 6 hr period: 25.6 ± 0.58, 25.8 0./9, 25.6 ± 0.50, 

25.5 ± 0.32, 25.6 ± 0.28, 26.9 ± 1.04, 25.7 ± 0.74, 26.1 2.42, 
25.4 ± 1.31, and 25.3 ± 0.68. These control excretion rates have 
not varied by more than 1 SD from the original mean, with 
frequent determinations, over a period of several months. 

The assay result is expressed as the ratio of the total excretion 
in cc/kg in excess of the mean control excretion for the same period 
of time to the sum of twice the standard deviations of the mean 


S(exper. excretion — mean control ex cretion) 
Diuretic ratio = s(2 X SD of mean control excretions) 

Tliis expression permits one to determine whether or not the in¬ 
crease of urine flow caused by a substance is significant, and it 
gives some indication as to the magnitude of the significance. 

Modification.-To compare an unknown substance with a stam - 
ard subS^ance (urea or Salyrgan). the procedure 
bv administering the unknown parenterally at the time ot ad 
ministration of the water load, or dissolved in the water load, t 
TanTmals and the standard administered in tlie same manner o 
3 ou the same day. Two or 3 

suffice for au as.say. The assay resu t is 
mponse, which is the average excretion in (( /k^ 
control mean excretion for each dog 

Diuretic response — 

S(exper. excretion - nu^a n conUoUxcrej^^ 

number of dogs 


and the diuretic potency is expressed as tiie ratio ol the diuretic 
responses for the unknown and for the standard. 

—This section was reviewed by Robert Gaunt. 


1. Lipschitz, W. L.; Hadidiaii, Z., and Kerposar, A.; J. Pharmacol. & E.xper. 

Therapy 79: 97, 1943. 

2. Little, J. M.: Unpublished observations. 

3. Little, J. M., and Cooper, C., Jr.: Federation Proc. 9: 296, 1950. 


ROBERT GAUNT, <Sj/rocuse TJjiiversily* 

Many attempts have been made to detect antidiuretic activity in 
blood, largely with the purpose of identifying circulating posterior 
pituitary antidiuretic hormone (ADH). Some successes have been 
reported (2, 14, 17-19, 22, 23). Failures to obtain any activity 
from blood have also been frequent (6, 10, 13, 15). There is prac¬ 
tically no agreement, however, among those who have found 
activity in blood as to whether the active agent is or is not ADH. 
The problem is inherently complex because of the low concentration 
of active material in blood, the lability of ADH in blood, and the 
possible presence of more than 1 antidiuretic agent and of inter¬ 
fering antagonist diuretic substances. If fractionation procedures 
are used, there is also the question of whether to separate materials 
of low or high molecular weight (1). 

Recently more promising results have been obtained by a method 
of ultimate simplicity, namely, the use as a test material of fresh 
whole blood, serum or plasma. This technique was probably first 
used successfully by Griffith and co-workers ’(9, 20). Although 
uniform results and interpretations have not resulted from such 
methods, antidiuretic activity in blood can be detected and in 
certain instances has been due almost certainly to the presence 
of ADH. 

The method described here resulted from the efforts of my 
colleagues, Birnie, Eversole and Jenkins (4, 5), to find a humoral 
antidiuretic substance (ADS) which might account for the failure 
of adrenalectomized rats to show normal water diuresis. In this 
work they found that normal rat serum was antidiuretic, and sub¬ 
sequent tests suggest, but peiliaps do not prove, that this activitv 
IS due to ADH. 

* Present address, Ciba PiiarmaceuUcal Products, Inc., Summit, N. J. 



Rats are anesthetized with ether, the thorax is opened and 
blood drawn directly from the right heart. Similar results are 
obtained with blood drawn under sodium pentobarbital anesthesia 
or by direct cardiac puncture without anesthesia. The blood is 
immediately centrifuged and an attempt made to get the serum 
injected into test animals within 5 min from the time it is drawn. 
If plasma is desired the blood is collected in syringes containing 
minimal effective amounts of heparin. Some acti\’ity is lost within 
30 min and most or all of it is lost if decanted serum samples remain 
in a refrigerator overnight, but systematic studies of the rate of 
inactivation of the ADS have not been made. 

Groups of 6 male rats, approximating 200 g body weight, are 
deprived of food but allowed access to drinking water for 18 hr 
before hydration. They are kept in small individual metabolism 
cages throughout the experiments. Three hydrating doses of water, 
warmed to body temperature, are given at hourly intervals by 
stomach tube, t The water dose is 3 ml/100 sq cm of body surface.! 
Slessor (21) prefers to use 0.2 per cent NaCl as a hydrating fluid 
rather than water. 

One hour after administration of the 2d dose of water the urine 
volume is measured. Any rat is discarded which during the 2 hr 
hydration period has a urine volume 50 per cent more or less than 
the mean of the group. The rats with normal diuresis are then 
given intraperitoneal injections of 1 ml samples§ of the test 
material and a 3d dose of water is administered. Thereafter the 
urine output is measured at 30 min intervals for 3 hr. The water 
excreted during this period is calculated as the percentage of the 
total water given (in 3 doses) minus that excreted before injection 
of the test material. The percentage of water excreted a+ 90 min 
after the injection of the test material was adopted as the definitive 
figure, and in ordinary experiments there is little to be gained by 
continuing observations beyond that time. 

The preliminary hydration period, which in principle has been 
used previously by Heller and Urban (12), permits the exclusion of 
animals likely to give aberrant responses and, m our opinion and 

be u.sed with great facility. ■Rr>nA<-lif>t (Vt bv the formula; S = 9.1 X 

: The surface area is cal^nulated as per Bene ^ ,method of calcu- 

B'V^ in which S is the response to water in rats of conspicuously 

lating water ^ A, a* these i rats of standard sizes are 

different sizes weig^t would certainly work as well. 

Guisbuls 0?) u», i ml /lOO g body weight, and for obvious reasons 
that is probably a preferable procedure. 



that of Slessor (21) (although complete experimental proof is not 
available), adds considerably to the sensitivity of the test animals. 

Precautions— The precautions we have observed most carefully 
concerned: (a) prevention of loss of urine, by holding the animal 
over its metabolism cage, when stomach-tubing was done or 
injections made; {h) getting the blood samples injected quickly, i.e., 
before inactivation, and (c) using docile test animals under con¬ 
ditions of minimal disturbance. The last is best accomplished by 
keeping the animals in a quiet place and having the stomach-tubing 
done by skilled technicians. We have used test animals only once, 
liut there is no established need for this precaution. 

Controls .—The well known daily variability in diuretic response 
of rats to water requires that controls be run simultaneously with 
test animals. Controls were given injections of normal saline solu¬ 
tion, with serum samples aged overnight and hence inactive, or 
of the fresh serum of hypophysectomized rats, such serum also being 
free from antidiuretic activity. 

Type of results obtained .—Fresh serum or plasma of rats tested in 
this Avay shows an antidiuretic activity roughly approximating 
that of 1 mU of Pitressin. It can readily be detected that there are 
differences in the amounts of blood ADS in various conditions (e.g., 
adrenal insufficiency, in which serum antidiuretic actityty is in¬ 
creased, or after hypophysectomy, when it is absent). The number 
of test animals required to reveal such differences will obviously 
vary with their magnitude. Usually 2 runs with 6 animals each 
have been made for use with statistical methods not involving 
variance analysis. Unfortunately, the activity of blood samples 
has not been carefully standardized against Pitressin, nor have 
adequate dose-response data been obtained. For that reason the 
method is spoken of here as one for the “detection” rather than for 
the “assay” of ADS. Some variation in the procedure, according 
to Dicker and Ginsburg (7), may reveal the presence of a stable 
antidiuretic agent which is a product of clotting; by following the 
method of Birnie et al. (5), however, Heller and Ginsburg (11) in 
the same laboratory have confirmed the presence of a labile ADS 
in nornial blood. They added the significant observation that in 
rats, mice and sheep, jugular serum contained much more anti- 
diuretic activity than carotid serum, as might be expected if the 
material was of pituitary origin. Therefore, if cardiac blood is used 
that from the right side of the heart should be more potent than 
that from the left. Other evidence concerning the possible identitv 
of the material m rat blood with ADH has been cited elsewhere 

Blood 0 / other species tested in ra(s.—Tliis method will reveal 


\'ariations in antidiuretic activity in the blood of various species 
(5, 11), including man (16, 21), Lloyd and Lobotsky (16) think, 
however, that there may be differences between the substances 
present in human and rat blood; perhaps they were dealing with 
the stable ADS of Dicker and Ginsburg. 

Use of whole blood .—Strong evidence has been provided by the 
van Dyke group (1) that ADH can be detected in dog blood. They 
observed antidiuretic activity in 10 ml samples of whole blood 
drawn from the jugular veins of dogs whose osmoreceptors had 
been stimulated by intracarotid injections of hypertonic sodium 
chloride. The test animals were dogs with diabetes insipidus. Un¬ 
fortunately, such test animals could hardly be used for routine 
clinical work. 

Comment by J. Mmwell Little 

Dr. Gaunt is to be congratulated on describing this procedure as a means 
of detection of antidiuretic activity rather than suggesting that it might be 
an assay method. 

Comment by Dr. Gaunt 

It is not yet an assay method, although we hope to make it one. The re¬ 
sults cannot be expressed in unitary quantitative terms, but it can be 
fairly said that one blood contains more or less activity than another, .^d, 
of course, the nature of the substance(s) in blood which has this antidiu¬ 
retic activity has not been established by direct means. 


2 . 


1 . 

Ames, R.; Moore, D. H., and Van Dyke, II. B.; Endocrinology 46; 216, 

Anselmino, K. J., and Hoffman, F.; Arch. Gynak. 147: 597, 1931. 

V n • vHnl Kn.p.rnetics (Washington, D. C.: Carnegie Institu- 


5 . 

6 . 


8 . 


10 . 

11 . 

12 . 







18. Marx, H., and Schneider, K.; Arch, exper. Path. u. Pharmakol. 176. 

24, 1934. 

19. Melville, K. I.: J. Exper. Med. 65: 415, 1937. . r 

20. Pendergrass, E. P.; Modes, P. J., and Grifl&th, J. Q., Jr.. Ain. J. Roent¬ 

genol. 46: 673, 1941. 

21. Slessor, A.: J. C. in. Endocrinol. 11: 700, 1951. 

22. Theobald, G. W.: Clin. Sc. 1: 225, 1934. 

23. Walker, A. M.: Am. J. Phj^siol. 127: 519, 1939. 


HENRY K. SCHOCH and AUGUSTO A. CAMARA, University of Michigan 

Urinary creatinine of a person subsisting on a creatinine-free 
diet is derived solely from muscle creatine and phosphocreatine. 
A relatively constant amount is excreted daily; this quantity is 
largely a reflection of muscle mass and not of muscular activity. 
Exogenous creatinine begins to be secreted by the renal tubules at 
plasma concentrations as low as 1.8 mg per cent (10). Creatinine 
produced in endogenous muscle metabolism, on the other hand, 
is believed not to be secreted at similar plasma concentrations. 
Neither exogenous nor endogenous creatinine is reabsorbed by the 
tubules. The available evidence suggests that both are freely fil- 
trable at the glomerulus. Thus it would seem that calculation of 
the renal clearance of endogenous creatinine would afford a meas¬ 
ure of glomerular filtration rate which should closely approximate 
the inulin clearance. Various investigators have obtained results 
in normal adult subjects which tend to support this view (6, 8, 
12, 15). Average endogenous creatinine/inulin clearance ratios were 
found to be 0.99-1.03, despite the fact that different technical 
methods were used. 


a) Dietary preparation .—When the clearance is to be determined 
over a 24 hr period, the errors introduced by exogenous creatinine 
are largely eliminated by use of a creatinine-free diet for the 2 days 
immediately preceding the test as well as during the test period 
itself. For practical purposes, the 40 g protein diet advocated by 
Addis (1) for patients with renal disease is an adequate substitute. 
It should provide a maximal daily meat intake of 60 S- 
shorter clearance periods (2-6 hr), use of a creatimne-free breakfast 
should suffice (2). The test should be scheduled for the morning 

hours immediately following the breakfast. 

h) Blood specimen.—k specimen for 
tion is obtained in the forenoon of the day of the test. In the case 
of ?he clinic patient who has completed a 24 hr collection period 
tt prior to Ws return visit, a blood ^Peci-o is obtained ^ soon 
after completion of the collection as is feasible, ith the dietary 

u- A- ■r.r. flip tPrm “endogenous creatinine clearance” will be used to 




precautions mentioned, significant variation in the level of serum 
creatinine from day to day does not occur. 

Hemolysis of the specimen or prolonged contact of red blood 
cells with the serum will result in falsely high values for serum 
creatinine concentration as the result of passage of noncreatinine 
chromogen from the cells to the serum. Care should be taken to use 
dry syringes, needles and containers and to assure gentle transfer 
of blood from syringe to container. Separation of serum from the 
clot after centrifugation should be completed within 1 hr. t 

c) Determination of serum creatinine concentration .—The method 
used is a modification of that described by Bonsnes and Taussky 
(4). A protein-free filtrate is prepared from 3.0 ml of serum by 
adding 21.0 ml of distilled water, 3.0 ml of 10 per cent sodium 
tungstate solution and 3.0 ml of 0.66N sulfuric acid in the order 
listed. Thorough mixing is recommended after addition of each 
reagent. The mixture is allowed to stand 20 min and then is filtered 
through no. 1 Whatman filter paper. The filtrate appearing in the 
1st 10 min is refiltered through the same paper, the collection being 
made in a 2d, clean test tube. This latter procedure assures a clear 

Duplicate 5.0 ml aliquots of the filtrate are transferred to color- 
irneter tubes, J and 2.0 ml of 0.04M picric acid solution is added 
with mixing. A blank is prepared by adding 2.0 ml of the picric 
acid solution to 5.0 ml of distilled water. To each colorimeter tube, 
including the blank, there is then added 2.0 ml. of 0.75N sodium 
hydroxide solution (which should be prepared with an accuracy 
of ±5 per cent as determined by standardization). The color 
which develops in each tube is then read in the colorimeter exactly 
20 min after addition of the sodium hydroxide solution to the tube 
in question. Using a Coleman Universal or Junior spectrophotom- 

Sives most accurate results. (For 
the Evelyn photoelectric colorimeter, filter #520 is inserted and the 
machine set for aperture 6.) 

Optical density values are compared with a standard curve 
Rowing optical density plotted against gamma of creatinine/5 ml 
This curve is prepared from 5.0 ml samples of standard solutions 
cont^l, 2, 3, 4, 5, 6, 7, 8, 9, 10, 20, 30, 40 and 50 gamma of 

sejaratlon^of centrifugation allows prompt 

1 large^tabTe'Sme'oVsp^^^^^^^ additional ai 

occasionally'LO^mf)^of^St\atTTs^i!l?d^^ TbSilfd^ content, only 3.0 or 2.0 ml (or 
measured amounts such as to produce a volume \ accurately 

picric acid and sodium hydroxid^e Sutions before addition of the 

IS made in calculating thLerum cSceStiom correction for this dilution 



creatiriine/5 ml, treated in the fashion described above wth the 
fully developed color being read in exactly 20 min. 

In our experience such a standard curve remains stable for sev¬ 
eral years as determined by frequent interval checking with stand¬ 
ard solutions. The Jaffe color does not adhere to Beer’s law, and it 
is important to dilute the serum (and urine) specimens such that 
their optical density values will fall on that portion of the curve 
which most closely approximates Beer’s law. In the standard curve 
described here, this corresponds to the range of 5-20 gamma of 
creatinine/5 ml. Sirota (14) has observed definite changes in the 
standard curve with different environmental temperatures and 
occasionally with a newly standardized sodium hydroxide solu¬ 
tion. As a consequence, he routinely uses 5 standard solutions with 
each series of determinations and dilutes urines and plasmas such 
that the expectetl values fall somewhere on the 1st half of the 
curve, but below 85 per cent transmission. When the Bonsnes- 
Taussky proportions are used, the color is stable for 15 min after 
full color development, as noted by Sirota (14), and a delay in 
reading of up to 15 min will thus not invalidate the observations. 

d) Uvinc s'pcciifncn .—If a 24 hr clearance is desiied, mine is col¬ 
lected by spontaneous voiding over an accurately timed 24 hr 
period, the specimen being preserved with 10 ml. of toluene 
Analysis should be performed promptly following completion ot 
nr.ilapUnn For shorter periods, spontaneous voiding may be 


e) Determination of urine 
the urine are made such tha 

IM i - ,1_ IT Ko f»Ynp<^t, 



Corrected clearance (in 1/24 hr) = 

24 hr urinary excretion (mg) 
eenun creatinine concen. (in mg/1) 



observed surface area (in sq m) 

If shorter collection periods are used, the calculations are made 
with a similar formula substituting the urinary excretion in 
mg/min for total 24 hr excretion and the serum level in mg/ml for 
mg/1, the final clearance being expressed as ml/min rather than 
as 1/24 hr. 

Noi'mal values .—We have routinely employed the 24 hr deter¬ 
mination of clearance for reasons discussed elsewhere (7). After 2 
days on a creatinine-free diet, the normal values for both males and 
females, corrected to a standard surface area of 1.73 sq m, average 
137 1/24 hr. The normal range is approximately 85-115 per cent 
of this figure. 

Sirota, Baldwin and Villarreal (13) found a mean value of 111 
ml/min for normal males on an unrestricted diet, utilizing a 4 hr 
collection period (8:00 a.m. to noon). Addis and co-workers (2), 
utilizing a similar collection period (7:00 a.m. to noon), obtained 
a mean value of 122 ml/min for normal males who had eaten a 
creatinine-free breakfast. The latter value is corrected to a stand¬ 
ard body weight rather than to a standard surface area. 

Limitations .—The unanswered problems regarding the nature 
of the various plasma chromogen substances which yield color with 
alkaline picrate place a serious limitation on the routine use of the 
endogenous creatinine clearance in studies which require accurate 
measurement of glomerular filtration rate. Some 75 per cent or more 
of the total plasma chromogen represents true creatinine in normal 
subjects (9, 11), while less than 5 per cent of urinary chromogen is 
other than true creatinine. Increase in noncreatinine chromogen 
content of the plasma, as occurs in cardiac disease (3, 13), will give 
falsely low clearances. Tubular excretion of creatinine, on the other 
hand, is known to occur in renal disease (5, 6) and tends to sWe 
clearance \'alues significantly higher than the true glomerular fil¬ 
tration rate. This test cannot, then, be considered a ^^alid investiga- 
tional technique in normal subjects or in individuals with disease 
yth the possible exception of the circumstance in which sufficient 
determinations of simultaneous inulin and endogenous creatinine 

rirrortl,^ ^leulation of filtration 

rate iiom tiie creatinine clearance. 

rate^ho^rV^' a dinfcoi measure of glomerular filtration 
late, hoive\ei the endogenous creatinine clearance is distinctlv 
superior to either the inulin or the urea clearance. Diurnal vari/- 

result from cha‘"“'’ “ g'omeriilar filtration rate which 

fiom changes in posture, variation in intra-abdominal pres- 



sure or the intake of food need not be considered when collections 
are made over a complete metabolic time period such as the 24 hr 
cycle (7). Reproducibility of results is surprisingly good, the limit 
of variability in 24 hr clearances usually being less than 5 per cent. 
The fact that the endogenous creatinine clearance overestimates 
glomerular filtration rate by as much as 10 per cent when renal 
function is reduced to Vs of normal (6), and by a progressively 
greater amount with further decrease in function, need not in¬ 
validate the determination as a useful method whereby the course 
of the patient’s disease may be observed. In actuality, the absolute 
differences between true glomerular filtration rate and endogenous 
creatinine clearance, even in patients with advanced renal disease, 
are so slight as to possess little clinical significance. 

Note. —This section was reviewed by Jonas H. Sirota. 

Comment hy A. C. Corcoran 

This article is intended for the attention of those in clinical practice who 
find that urea clearance is not a satisfactory index of glomerular filtration 
rate under the conditions in which these measurements are so often made. 
Where such is the case, an alternative procedure which concentrates its 
pitfalls in the laboratory and in the economy of the body rather than iin 
the ward and ward attendant would be welcome. The author and the 
reviewer have alike had wide experience with the clinical use of endogenous 
creatinine clearance and they conservatively assess its value. 


1. Addis, T.: Glomerular Nephritis: Diagnosis and Treatment (New York: 

2 Add\rT^^ ai: The relation between protein consumption and 

variations of the endogenous creatinine clearance in normal individuals, 

3. K.: ^ “c-tinine” c.ear- 

4 . BornVrw;;7nd^uu::*y: 

5 BrTt'ana’'KSo:xrS;i of eodogenous creatinine, 

a. bS 7, ana Safa: “fpj'ijnal tdearance of endogenous 'Vreatinine- 

j"”“"-/L'‘”\'L‘"TheLentv’dour\ourly endogenous creatinine elear^ 

' ■ “aTras a ctieal mei^e of tire funet.onal state of the krdneys, J. Bab. 

8. H^e? Renal’ exLtion of creatinine in man, Federation Proc. 

9. Hare^R. S^’^and Hare, K.: Determination of creatinine in blood and urine. 

Federation f> 8 . B|,,.ation and secretion of exogenous 

10. Jolliffe, N., and Chasts, ^ , 933 . 

creatinine in man, Am. J. 1 hv. 


11. Miller, B. F., and Dubos, R.: Determination by a specific, enzymatic 

method of the creatinine content of blood and urine from normals and 
nephritic individuals, J. Biol. Chem. 121: 457, 1937. 

12. Miller, B. F., and Winkler, A. W.: The renal excretion of endogenous 

creatinine in man: Comparison with exogenous creatinine and inulin, 
J. Clin. Invest. 17: 31, 1938. 

13. Sirota, J. H.; Baldwnn, D. S., and Villarreal, H.: Diurnal variation of 

renal function in man, J. Clin. Invest. 29: 157, 1950. 

14. Sirota, J. H.: Personal communication. 

15. Steinitz, K., and Tiirkand, H.: The determination of the glomerular filtra¬ 

tion rate by the endogenous creatinine clearance, J. Clin. Invest. 19: 



KARL H. BEYER and LEMUEL D. WRIGHT. Sharp and Dohme, Inc., 

Tl’est Point, Pa. 

Renal tubular participation in the elaboration of nrine consists 
in modifying glomerular ultrafiltrate by a variety of secretory and 
reabsorptive processes. The energy available for the enzymatic 
mechanisms on which these functions depend seems to be allocated 
and committed, as it were by quanta, to functions some of which 
are highly specific and some, e.g., growth and repair more general 
(1). Grouping of functional and metabolic processes is evidenced 
by the mutual competition of some compounds (e.g., PAH and 
penicillin) for secretory mechanisms (2) and others (glucose and 
xylose, arginine and lysine) for reabsorptive pathways (2, 4, 13). 
The energy requirements of other processes are seemingly inde¬ 
pendent, as indicated by the lack of competition for concurrent 
secretion of PAH and N^-methylnicotinamide or of competition 
for reabsorption between essential monoamino and diamino acids 

(3, 4). 

Renal disease, renotropic agents and nephrotoxins may differen¬ 
tially alter specific functions. With recognition of selective altera¬ 
tions of functions has come the development of methods that 
yield specific information relating to maximal tubular secretory 
capacity (Diodrast, PAH) or reabsorptive capacity (glucose). 
Measurement of renal tubular reabsorptive capacity for arginine 
and lysine can contribute to a still more precise analysis of tubular 
disease and malfunction. The reabsorptive pathways for these 2 
amino acids are at least in part the same, as shown by mutual 
competition for reabsorption. Further, their reabsorptive maxima 
are better defined than those of most other ammo acids (5, 11, 20 . 

Of these 2 amino acids, natural arginine is commercially avail¬ 
able at a reasonable cost. Its microbiologic determination m body 
fluids is not difficult. Microbiologic assays are more specific than 
determinations based on detection of ammo or carboxyl p-oups^ 
Since arginine is substantially nontoxic when administered laig 
doses to^dogs, determination of its maximal rate of reabsorptio 
fTm) can be applied to the study of aberrations of tubular function 
which result in hyperaminoaciduria whether of congeni a , reno 

of n,ost other reabsorptive oorxima, it is 




assumed tliat arginine is not secreted by the tubules and that Tm 
represents the difference between the amount of the ammo acid 
that is presented to the tubules in glomerular filtrate and that 
which is concurrently excreted in the urine. The plasma concen¬ 
trations must be sufficiently high during Tm determinations that 
the reabsorptive capacity (Tm) and not the load {GF • P) presented 
for reabsorption limits that rate. An arginine plasma concentra¬ 
tion of about 20 mg/100 ml (200 7 /ml) has seemed satisfactory 
for this purpose. 

The maximal reabsorptive capacity for arginine may be expressed 
as a formula 

Tm (mg/min) = (P-GF) — UV 

where P = plasma concentration of arginine in mg/ml; GF = 
creatinine clearance or glomerular filtration rate in ml/min; 

TABLE 1.— Protocol for Arginine T.m Estimation 
Dog 370, 12.4 kg, surf, area = 0.60 sq m, Exper. 9196 

Urine Wash- Vol, r'R'+ 

Time, Elapsed, Sam- Vol, out Ml/- Approx. Actual 

Hr: Min Min ple Ml* Dil. Min V Dil.J Dil.§ 

Postabsorptive condition. Fasted overnight. 

0:00 650 ml HjO by stomach tube 

1:00 400 ml H 2 O by stomach tube 

1:25 L(+)-arginine venoclysis begun; 5 mg/kg/min in normal saline at 
3 ml/min'l 

1:26 Creatinine 2.5 g subcutaneously (25 ml of 10% sol.) 

1:30 Arginine intravenous single “priming dose” = 3.0 mg/kg (2 ml of 
1.86% sol.) 




























u ui j j * ^ . 1 “ . 8i.eriie water lor Diaaaer wasii-out, wnich was followed by insufflation 
of the bladder with air to assure complete emptying. 

t Estimated glomerular filtration rate (OF) = 60 ml/min. 

I o“v 1 is .rounded off (9.0) for analytical convenience. 

'Tu- . 1 ® ® .urine was used according to the procedure outlined) = 23..58 

ihis IS the actual dilution used to convert mg of creatinine/tube to mg/100 ml in Table 

tioM^for analysir^''^ concentrations would approach plasma concentra- 

1 " arginine in normal saline is sufficient for the “priming dose” and veno¬ 

clysis for this experiment using a 12.4 kg dog. 

UV = amount of arginine excreted/min, where U = arginine 
concentration in urine in mg/ml and V = urine flow in ml/min. 


Determination oj renal clearance .—A protocol for determination 
ot renal clearances from which the arginine Tm is calculated is 



presented in Table 1, Table 2 summarizes the calculation of creati¬ 
nine and arginine clearances and the final expression of Tm. 

In general, dogs are fasted overnight in order to avoid the dynamic 
action of foodstuff and to enhance the absorption of materials 
administered by gavage. The animals are given sufficient water to 
assure adequate urine flow in order to minimize errors due to re¬ 
tention of small amounts of urine or sterile water with which the 

TABLE 2.— Calculation of Ckeatinine Cleabance and Arginine Tm 
D og 370, 12.4 kg, surf, area = 0.60 sq m, Exper. 9196 



Galv. Reading 

cate Av. 





















































100 Ml 




















Sq M 























nnal 2-fold 


them into a more critical Pa/t o^ inappropriate for arginine. The y/m\ 

t The system of dilutions used for creatinine Table 3. , . 

+ Arginiue p j surface area of 1.0 sq m 

by^dlTding actual amount reab.sorbed by actual surface area of the dog (0.60 sq m). 

(Amt filtered — UV)^ This has been corrected to a 

w ;«ivB^Vipd The “priming” and venoclysis of arginine are to 

Sn qS kf-d to maintain® high plasma concentrations of 

it^^f t\eTt:“thi |-ed 

Im^nHrf .m-giniim reabsorbed per unit time is the difference 



between the amount filtered and that which is excreted, it is 
sary to determine the simultaneous glomerular filtration rate which 
is equivalent to creatinine clearance in the dog (10, 14) and to 
mannitol (18) or inulin clearance in man (16). 

Creatinine deterininaHons.— are carried out on a protein- 

free filtrate prepared by the alkaline picrate method of Folin and 
Wu (7). 

A. jMethod: a) Preparation of protein-free pla.sma filtrate, 
mix 1.0 ml plasma, 0.5 ml 0.G7N sulfuric acid, 0.5 ml 10 per cent 
sodium tungstate and 18 ml distilled water; let stand for 10 min 
and filter. 

h) Dilution of urine for analysis: 

estimated clearance (ml/min)_ 

urine vol (ml/min) X wash-out dilution 

c) Preparation of alkaline picrate solution: immediately before 
use, mix 1.0 ml 10 per cent sodium hydroxide, 5.0 ml saturated 
picric acid and G ml distilled water. 

d) Pipet 5.0 ml plasma filtrate from (a) or 0.5 ml diluted urine 
plus 4.5 ml distilled water (/>) into colorimeter tube. Add 10 ml 
of alkaline picrate solution (c) and let stand 15 min. Read in an 
Evelyn colorimeter using a §520 filter, 10 ml aperture against a 
blank tube containing 5.0 ml distilled water and 10 ml alkaline 
picrate solution. 

B. Calculations: a) Plasma (mg/100 ml) = mg/tube (read 
from curve) X 400 

b) Urine (mg/ml) = mg/tube (read from curve) X 2 X wash¬ 
out X urine dilution 

s r'^ / ^ \ urine cone. X urine vol 

c) Clearance (ml/min) =--- 

plasma cone. 

\\ ith an Evelyn photoelectric colorimeter using a §520 filter, 
creatinine concentration in plasma and urine is conveniently 
estimated. 4 he galvanometer deflections for the various samples 
are read against a reagent blank set at 100 (air blank = 88°) 
and compared to a standard curve for estimation of absolute 
values. Creatinine data corresponding to the protocol of the 
experiment given in Table 1 are set forth in Table 2. 

Arginine determination. Ihis is performed on protein-free 
filtrates prepared with tungstic acid according to the method of 
Dunn et al. (G). For the tungstate precipitation of plasma, mix 2 
ml pla.sma, 8 ml water, 1 ml Vs N sulfuric acid and 1 ml of 10 per 
cent sodium tungstate; let stand for 2 min, centrifuge, then neutral¬ 
ize the centrifugate. 

Ihe microbiologic estimation of arginine according to the 
method of btokes et al. (18), using his basal medium and Strevlo- 



coccus Jaecalis R as the organism, is recommended. Stoke’s l)asal 
medium contains: 


100 mg 


100 mg 


100 mg 

L( —)-cystine 

100 mg 


100 mg 


200 mg 

L( —)-tyrosine 

100 mg 


100 mg 

DL-glutaniic acid 

100 mg 


100 mg 


100 mg 

DL-aspartic acid 

4 00 mg 


50 mg 


100 mg 


100 mg 

L( —)-proline 

100 mg 

L( — )-hydroxyproline 

100 mg 


100 mg 


100 mg 


5 g 

sodium acetate (anhy¬ 

drous) 3 K 

adenine 5 m^ 

uracil 5 mg 

guanine 5 mg 

pantothenic acid 100 7 

riboflavin 100 7 

thiamine hydrochloride 100 7 

nicotinic acid 100 7 

pyridoxamine 200 7 

p-aminobenzoic acid 20 7 

biotin 0. 1 7 

folic acid 1.0 7 

salts A 

Iv,HP ()4 250 mg 

KH 2 P ()4 250 mg 

salts B 

MgS 04 - 71 l 20 100 mg 

NaCl 5 mg 


Adjust to pll 0.8and add distilled water to 250 ml. 

0 mg 
5 mg 

TABld'i 3.— Assay Protocol for Microbio logic Estimation of Arginine’ 


Standard Curve 

Unknown Samples 


Sol., Ml 

Final Conc. 
INE. 7/10 Ml 



Ml oe Dil. 


Ml or Oil. 





























1 I 













* Five ml of double strength medium is added to tubw ^ sfr?/oecoIL R cells ob- 

LfclASo. Cl. in .0 .i-n. .h. or,.- 

inal volume of sterile saline. 

\ Standard curve for L(+)-arginine is set up ^''jth each expen- 
inent as indicated in Table 3. After 18 hr of incubat.o.i at C, 



turbidimetric estimations of growth are performed using a Klett- 
Summerson photoelectric colorimeter, a filter and a galvanom¬ 
eter zero setting against water as a reference medium. The 
amounts of arginine present in plasma and urine are estimated by 
comparison of galvanometer readings of the unknown samples with 
the standard curve run concurrently. Calculations of arginine 
clearance and Tm values for the experiment presented in Table I 
are summarized in Table 2. 

Co 77 inient—l\usso ei al. (12) analyzed the result of arginine Tm 
values of dogs employed in clearance studies in this laboratory 
over the years 1946, 1947 and 1949. It was found that the difference 
between Tm values was not significant but that the mean arginine 
Tm for 1949 was lower than that for 1946 and 1947. Even so, the 
mean value and standard of a single determination for any dog for 
any one occasion was 12.9 ± 1.7 mg/min/sq m of surface area. 

TABLE 4.—Reproducibility of Arginine Tm o.v Different Days 
Dog 370, 12.4 kg, surf, area = 0.60 sq m 













Mg/Min/Sq M 

1st Exper. 











2d Exper. 











The 95 per cent confidence limits were 12.9 ± 3.3 mg/min/sa m 

( 8 ). ^ 

Table 4 illustrates the reproducibility of successive Tm values 
based on 10 min clearance periods. The constancy of arginine Tm 
determinations on different days for the same animal is shown 
in Table 4 by the data from 2 experiments performed several days 
apart on the same dog. This reproducibility of arginine Tm ob¬ 
tained despite a considerable variation in glomerular filtration on 
the 2 occasions. 

We have used tlie mierobiologic estimations of amino acids for 
clearance determinations because of their greater speciBcity. 

itts first described the Tm for arginine using the more conven¬ 
tional ninhydrin reaction, which reacts in general with alpha- 
ammo acids (9), However, under conditions where the values for 
arginme 1 m are to be given greatest interpretive weight, the micro- 
biologic estimation inherently should be the most critical In the 






hands of experienced laboratory workers, quantitative micro¬ 
biological assays are as reliable as are chemical determinations. 


1. Beyer, K. H.; Functional characteristics of renal transport mechanisms, 
J. Pharmacol. & Exper. Therapy 99; 227-280, August, 1950. 

2a. Beyer, K. H., et al.\ The prolongation of penicillin retention in the body 
by means of para-aminohippuric acid. Science 100: 107-108, Aug. 4, 

2b. Beyer, K. H., et al.\ The enhancement of the physiological economy of 
penicillin in dogs by the simultaneous administration of para-amino- 
hippuric acid, J. Pharmacol. & Exper. Therapy 82; 310-323, Decem¬ 
ber, 1944. 

Beyer, K. H., et at.: The renal tubular elimination of N‘-methylnicotin- 
amide. Am. J. Physiol. 160; 311-320, February, 1950. 

Beyer, K. H., et ai: Renal clearance of essential amino acids: Their com¬ 
petition for reabsorption by the renal tubules. Am. J. Physiol. 151: 
202-210, November, 1947. 

Beyer, K. H., et al: The renal clearance of essential amino acids: TryiJto- 
phane, leucine, isoleucine and valine. Am. J. Physiol. 146: 330-335, 
June, 1946. 

Dunn, M. S., et al: Investigations of amino acids, peptides, and proteins: 
Determination of apparent free tryptophane in blood by microbiologi¬ 
cal method, J. Biol. Chem. 157: 387-394, January, 1945. 

Folin, O., and Wu, H.: A system of blood analysis, J. Biol. Chem. 
81—110 May, 1919. 

Ganguli, M.: A note on nested sampling, Sankhya, Indian J. Statistics 

5 (pt. 4): 449-452, 1941. . r .,1 

Pitts R F • A comparison of the renal reabsorptive processes for several 

amino acids. Am. J. Physiol. 140: 535-547, January 1944. 

Richards, A. N.; Westfall, B. B., and Bott, P. A.: 

inulin, creatinine and xylose in normal dogs, Proc. . P 

RU^ r^VaiSl-Le of -ntial anrino ac^s. Threonine 
and phenylalanine, Proc. Soc. Exper. Biol. & Med. 65. 215 217, June, 

it • f'iminpra J L and Gass, S. R.: Statistical analysis of 
^rallLra;me“^Th:dog, Federation Proc. 10: n^Mie^ 
Shannon J A.; The tubular reabsorption of xylose m the nor g, 

Am. J.’Physiol. 122: 775-781 June 1938.^^ ^ 

^^te'^dor vt'’l^i1'filt;atio\rL^ s^crebon of'exogenous creatinine. Am. 

Tm 111 the normal T^e^renal tubular reabsorption of glucose 

15b. Shannon, J. A., and Iisher, i e - 05^774 June, 1938. 

in the normal dog. Am J. Phys 1^2 765 of’inulin, xylose and urea 

Shannon, J. A., and Smith, H.^^1 he excret 14 . 393 - 4 OI, July, 

by normal and phlonzinized man, J. ciin. mvc 

24; 388-404, May, 1945. 


8 . 












18 Smith, W. W.; Finkelstein, N., and Smith, H W.; Renal excretion of 
hexitols (sorbitol, mannitol and dulcitol) and their derivatives (sorbit- 
an, iso-inannide and sorbide) and of endogenous creatinine-like chro¬ 
mogen in dog and man, J. Biol. Chem. 135; 231-250, August, 1940. 

19. Stokes, J. L., et al.: Microbiological methods for the determination ot 

amino acids; II. A uniform assay for the ten essential amino acids, J. 
Biol. Chem. 160; 35-49, September, 1945. 

20. Wright, L. D., et al.: The renal clearance of essential amino acids; argin¬ 

ine, histidine, lysine and methionine, Am. J. Physiol. 149; 130—134, 
April, 1947. 


JOHN V. TAGGART, Columbia University, and ROY P, FORSTER, 

Dartmouth College 

A GROWING interest in the metabolic aspects of cellular transport 
has led to the development of a variety of in vitro techniques which 
are particularly useful in studying renal transport mechanisms. The 
procedures to be described have certain features in common and 
similar advantages and disadvantages. They represent outgrowths 
of the method introduced by Robert Chambers and his co-workers 
in which the accumulation of phenol red was observed in tissue- 
cultured cystic explants of embryo chick mesonephros. 

The absence of glomerular filtration in any in vitro preparation 
of kidney limits observations to those on tubular excretion; no 
satisfactory methods exist for studying tubular reabsorption in 
vitro. Certain of the excretory mechanisms are known to be active 
processes capable of establishing large concentration gradients 
across the tubular epithelium. Consequently, satisfactory quantita¬ 
tive results can be obtained in vitro with small tissue samples. 

The in vitro techniques are not intended to replace in any sense 
the well established experimental procedures for renal studies in 
the intact animal. Their principal usefulness lies in facilitating 
biochemical observations on transport mechanisms. The compo¬ 
sition of the ambient fluid can be rigidly controlled and can be 
varied over a much wider range than is possible in the intact animal. 
Various chemical agents and enzyme inhibitors, too toxic for use m 
the intact animal, can be examined for their specific effects on 
selected renal mechanisms. The extrarenal effects of these agents 
which may alter kidney function indirectly are excluded, hmally, 
the technical simplicity of the procedures makes possible a much 
larger number of experimental observations than can be con\en- 
iently obtained by the clearance technique. 

I. Phenol Red Transport in Isolated Fish Tubules 
and Frog Kidney Slices 

Lona segments of renal tubules can be leaclily isolated from the 
kidnev^s of many marine and fresh water fishes, h ish kidneys are 
particularly suitable because a very limited iiitertubiilar cement 




su})staiu’e permits the separation of individual tubules. Satis¬ 
factory preparations ha\'e been obtained from the common nintei 
flounder {Fseudoplmronectes americanus), killifish {Fundulus 
heteroclitus), sculpin (Myoxocrphalus oclodecimspinomH) and the 
fresh water hoi'iipout, catfish or bullhead {Ameiuru^ nebulosus). 
The flounder tul)ule is apparently undifferentiated, whereas that 
of Ameiuriis contains I'eadily distinguishable proximal and distal 
convoluted segments. 

When isolated fish tubules are incubjited in a dilute solution ot 
phenolsulfonphthalein (phenol led) in a balanced saline medium, 
concentration of the dye is apparent within the tubular lumina in 
5-10 min and increases in 30 60 min to approximately 100 times 
that in the medium. The principal advantage of the isolated tubule 
preparation ovei’ Chaml)er’s technique is in obviating the use of 
tissue culture methods. 


1. Isolated fish tubules. —Observations should be made only 
on freshly caught live specimens. The abdominal cavity is cut open 
and the kidney transferred to a beaker of the saline medium (with¬ 
out phenol red). Fiagments approximately 1 mm across are teased 
from the kidney with fine forceps and placed in individual Petri 
dishes containing 5 cc of a freshly prepared and actively oxygen¬ 
ated medium of the following composition: 

FOR MARINE FISHES: NaCl 0.783 per cent (134 mM), KCl 0.0186 
per cent (2.5 mM), CaClo 0.0167 per cent (1.5 mM), MgCl 2 0.00953 

per cent (1.0 mM), XaH 2 P 04 Ti 20 0.0069 per cent (0.5 mM), 
NaHCOs 0.126 per cent (15.0 mM), and phenol red 0.001 per cent 
(0.028 mM). 

FOR FRESH WATER FISHES: Same as for marine fishes, except that 
NaCl is reduced to 0.585 per cent (100 mM) and NaHCOs to 0.042 
per cent (5.0 mM). 

The pH of the medium is approximately 8.4, but satisfactory 
phenol red accumulation can be obser\'ed within the pH range of 
^d)—8.6. Although bicarbonate is not an essential component, the 
alkalinity which it imparts to the medium insures the red color of 
the dye and facilitates quantitative comparisons. Experiments are 
performed at room temperature. Continuous oxygenation is sup¬ 
plied to each Petri dish through a 2 in. 22 gauge hypodermic needle 
bent to fit over the rim. The initial teasing, together with the gentle 
agitation pi'ovided by the stream of oxygen bubbles, quickly 
separates the individual tubules. Fi-om time to time the Petri dish 
contammg the kidney fragments is placed on a microscope stage for 
examination with 50 100X magnification. The concentration of 



phenol red in the lumina is graded as follows: 0 indicates no 
accumulation, + a barely detectable concentration (5-fold), and 
-b-j- to -1- + + + successive increases to approximately 100-fold. 
Since all the tubules do not accumulate dye at an identical rate, 
the sample is graded on the basis of the most active segment 

Metabolic intermediates and inhibitors are dissoh^ed in the 
medium and the pH is readjusted to 8.4. In the usual experiment, 
7 Petri dishes are run simultaneously; 1 serves as the control and 
the remaining dishes contain various concentrations of the agent 
under stud 3 ^ Experiments are considered satisfactory only when 
the control shows -b-b-f-j- concentration of dye within 30-60 min. 

2. Frog kidney slices .—Frog kidney can be sliced freehand 
with a razor blade into thin longitudinal sections which are satis¬ 
factory for studying phenol red transport. The incubation medium 
is the same as that used for fresh water fishes except that 0.5 per 
cent glucose is added. A natural yellow pigmentation of frog kidney 
slices makes the interpretation of dye concentration somewhat 
more difficult than in the isolated tubule preparations. 

II. Phenol Red Transport in Guinea Pig Kidney Slices 

Beyer and his associates (2) have modified the method of Forster 
( 6 ) to permit the study of phenol red transport in slices of mamma¬ 
lian kidney. Guinea pig renal cortex is apparently the most satis¬ 
factory, but qualitatively similar results have been obtained with 
rabbit, mouse and rat kidney. In slices of mammalian kidney 
phenol red transport can be shown to occur only in the proximal 

convoluted tubules. 


A guinea pig is exsanguinated and the kidneys are removed to a 
freshly prepared and chilled modified Krebs-Ringer bicarbonate 
solution of the following composition: NaCl 0.85 per cent (144 m. ), 
KCl 0.043 per cent (5.75 mM), KH 2 PO 4 0.019 per 
MgS 04 - 7 H 20 0.036 per cent (1.44 mM), and NaHCOs 0.403 pei 
cent (48 mM). Thin slices of kidney cortex are sectioned freehand 
with a razor blade and incubated in 5 cc of medium through which 
5 ner cent C 02-95 per cent O 2 is bubbled continuously. I he final 
dH of the medium is 7.8. The temperature during incubation is 
maintained at 37 C in a thermostatically controlled glass chambei. 
Mter 5 min of incubation, phenol red is added to the medium m a 
fnnl concentration of 4 mg per cent. After an additional 15 30 min 
t iucrare^moved, spld on a glass slide and exam.ned .mmed.- 



ately under the microscope at 30-60 X magnification. The accumu¬ 
lated dye appears as streaks of red clearly outlining the tubular 

lumina. , • u-u 

Studies in Warburg apparatus .—Studies with volatile inhibitors 

(e.g., HCN) can be performed in the Warburg apparatus. Beyer 
recommends that the tissues be exposed to inhibitors, particularly 
cyanide, for several minutes before addition of phenol red. 
Consequently, the kidney slices and medium are added to the 
main compartment of the vessel and the phenol red is tipped in 
from a side-arm after a 5 min equilibration period. The gas space 
is filled with 5 per cent C02-95 per cent O 2 . The slices are removed 
from the vessel after 15 min and examined in the usual manner. 
When measurements of respiration are desired, the medium is 
modified by replacement of NaHCOs by an equivalent amount 
of sodium phosphate buffer of pH 7.8, the C0?-02 mixture is re¬ 
placed by 100 per cent O 2 , and alkali is placed in the center well 
of the Warburg vessel. 

III. Transport of PAH, Diodrast and Phenol Red 
in Rabbit Kidney Slices 

The slice technique (6) has been modified by Cross and Taggart 
(4) to permit quantitative studies on the tubular transport of 
compounds other than phenol red. 


Experiments are performed in the conventional Warburg appa¬ 
ratus. For studies on PAH transport, the main chamber of each 
vessel generally contains 1.18 cc of 0.3M NaCl, 0.27 cc of 0.1 M 
KCl, 0.3 cc of O.IM Na acetate, 0.2 cc of 0.1 IM Na phosphate 
buffer of pH 7.4, 0.1 cc of 0.02M CaCb, 0.2 cc of O.OOIM Na PAH, 
and water to a final volume of 2.7 cc. When other compounds 
requiring neutralization are used, they are added as the sodium 
salts and the NaCl is reduced by an equivalent amount. The 
loaded vessels are chilled in crushed ice. 

A rabbit is exsanguinated and the kidneys are promptly trans¬ 
ferred to a beaker of ice-cold 0.13M NaCl-0.02M KCl. Slices of 
renal cortex 0.3-0.4 rnm thick are cut in a Stadie-Riggs microtome 
(9). After brief blotting on filter paper, the slices are weighed on a 
torsion balance and approximately 300 mg is added to each vessel 
Filter paper and 0.2 cc of 6N NaOH are placed in the center well’ 
the cups are mounted on manometers, and the gas space is filled 

in a 25 C water bath at 
a rate of 100 c/mm. After 15 min equilibration, the taps are closed 



and oxygen consumption is recorded during the following 30 min. 
Oxygen uptake is expressed as a QO 2 (cu mm of O 2 consumed/mg 
initial wet weight of tissue/hr). 

After 1 hr in the bath, the vessels are removed and promptly 
chilled in crushed ice. The slices are removed as soon as possible, 
blotted carefully on filter paper, weighed and transferred to a 
graduated cylinder to which are added 3 cc of 10 per cent tri¬ 
chloroacetic acid and water to a final volume of 10 cc. A 2 cc ali¬ 
quot of the final suspending medium is treated with trichloroacetic 
acid in a similar manner. The solutions are filtered after 20 min 

Fiq 1—Accumulation of PAH by rabbit kidney slices. 

and 5 cc aliquots of the filtrates are used for estimation of PAH by 
the method of Bratton and Marshall (3). The accumulation ot 
PAH in kidney slices is expressed as the ratio of PAH concentration 
in slices/medium (S/M), as shown in Figure 1. Recoveries of 

added PAH average about 98 per cent. ... .1 

The rate of PAH accumulation in rabbit kidney slices is grea y 
enhanced by presence of acetate in the suspending medium (4; 
aher oxidiL&e substrates, such as ">-bers of the c, ejd 
cycle, certain amino acids and the lonpr ehain tatty acids, are 

Strongly inhibitory in concentrations as low as O.OIM. 

For studies on I Hodrast Iramport, the commercial product is 



converted to the sodium salt by crystallization of the free acid 
and subsequent neutralization with NaOH. Each vessel contains 
0.2 cc of 0.0075M Na Diodrast in place of PAH. Cadmium fil¬ 
trates ( 8 ) of the slices and medium are used. The slices should be 
allowed to extract in the CdSO^—H 2 SO 4 solution for 20 min before 
the addition of NaOH. Diodrast is estimated by the iodometric 
method of Alpert (1). 

Quantitative studies with phenol red are somewhat less satis¬ 
factory in that the dye accumulated in the slices cannot be com¬ 
pletely recovered. The final concentration of phenol red in the 
medium is estimated directly by diluting an aliquot in 0.01 N 
NaOH and measuring the optical density at 550 myn. The total 
quantity and concentration in the slices are calculated on the 
basis of the difference between the amount of phenol red added 
and that reco^^ered in the medium. 

Note. —This section was reviewed by Karl H. Beyer. 

Comment by A. C. Corcoran 

The techniques here described provide a novel, dramatic and undoubt¬ 
edly informative approach to the study of the mechanisms of tubular 

1 . 

2 . 



0 . 


8 . 

10 . 


.41pert, L. K.: Rapid method for determination of diodrast-iodine in blood 
and urine, Bull. Johns Hopkins Hosp. 68: 522, 1941. 

Beyer, K. H.; Painter, R. H., and Wiebelhaus, V. D.: Enzymatic factors 
in renal tubular secretion of phenol red. Am. J. Physiol, iei; 259, 1950. 

Bratton, A. C., and Marshall, E. K., Jr.: A new coupling component for 
sulfanilamide determination, J. Biol. Chem. 128: 537, 1939. 

Cross, R. J., and Taggart, J. V.: Renal tubular transport: Accumulation 
of p-aminohippurate by rabbit kidney slices. Am. J. Physiol. 161: 181, 

Dearborn, E. H.: Inhibition of the appearance of phenol red in frog kid¬ 
ney tubules in vitro, Proc. Soc. Exper. Biol. & Med. 70: 105, 1949. 
Forster, R. P.: Use of thin kidney slices and isolated renal tubules for di¬ 
rect study of cellular transport kinetics, Science 108: 65, 1948. 

Forster, R. 1., and Taggart, J. V.: Use of isolated renal tubules for the 
examination of metabolic processes associated with active cellular 
transport, J. Cell. & Comp. Physiol. 36: 251, 1950. 

Smiith, H. W., ei al: The renal clearance of substituted hippuric acid 
derivatives and other aromatic acids, J. Clin. Invest. 24: 388 1945. 
Stadie, W- C., and Riggs, B. C.: Microtome for preparation of tissue 

1 metabolic studies of surviving tissues in vitro, J. Biol. Chem 

154; 687, 1944. 

^" 1 ^ Forster, R. P.; Renal tubular transport: Effect of 

on phenol red transport in 

the isolated tubules of the flounder. Am. J. Physiol 161- 167 1950 


A. J. DALTON, National Cancer Institute 

Early efforts in the development of methods of obtaining ultra- 
thin sections for observation of tissues with the electron microscope 
included use of wedge sections (25) and development of the high¬ 
speed microtome (11, 20). These efforts were moderately success¬ 
ful and certainly demonstrated the feasibility of thin sectioning for 
electron microscopy but involved a considerable element of chance. 
The highspeed microtome remains a useful mechanism for obtain¬ 
ing thin sections of excessively hard substances, but the adaptation 
of various models of microtomes normally used for routine his¬ 

tologic sectioning for use in ultrathin sectioning (1, 2, 9, 10, 14, 18, 
23) and introduction of the glass knife (16) has made tissue sec¬ 
tioning for electron microscopy practically a routine procedure in 
many laboratories. Although in this laboratory our experience has 
been confined almost exclusively to the thermal expansion method 
of sectioning following the n-butyl methacrylate embedding method 
(18), it has been demonstrated that with proper use of any one of 
these methods, sections satisfactory for electron micrography may 
be obtained (2, 9, 10, 18, 23). Thus the problem of thin sectioning, 
which many investigators undoubtedly felt would take many years 
to solve, is no longer a barrier to progress in the field. 

Fixation of tissues for electron microscopy, which until recently 
received but little attention, now looms as a major and complex 
problem, relatively difficult of solution. The criteria of classic 
cvtoloo-y for determining the adequacy of fixation are of some aid 
in this’^respect, e.g., comparison of structures m fixed preparations, 
such as nucleoli, nuclear membrane, mitochondria and cell mem- 
brane, with their counterparts in living cells. \et it is well knovn 
that materials apparently satisfying these criteria reasonably well 
under the light microscope, reveal relatively gross artefacts vhe 
examined with the electron microscopy There ™ 
obvious problem of how to determine the degree “ 

nrese)(-aLn of structures too small to be resolved with the light 

microscope. In the absence of any standard of 

ence a reasonable approach would appear to be that ot assumii g 
tentatively that a process of fixation which preserc'es microscopic 
"res\a!lcquately on exposure to the electron beam also pro- 


, 1—I’ortioii of Ji hepatic cell of a mouse illustrating results obtained 

following fixation by perfusion with the osmic, dichromate, lanthanum mix¬ 
ture. Nuclear detail is reasonably good and mitochondria and submicroscopic 
lain^lae are sharply delineated in the cj'toplasni. Approximately 4 800X 
(In Figs. 1-5, the iiorizontal line is the equivalent of 1 ju.) ' ' 

fi f^pight portion of a proximal tubuie of a mouse kidney 

hxed as the cel in higure 1. Nuclear contents, nuclear membrane and mito- 
cnondria show little evidence of distortion. Approximately 4,800X. 



mi-:tiioi)S of renal study 

serves siibmicroscopic structures satisfactorily under the same 

It has long been known that, with direct contact, osmic acid in 
solution or in vapor form fixes and preserves cytoplasmic com¬ 
ponents more (piickly and in more nearly the form and distribution 
they possess in living cells than any other of the fixatives of classic 
cytology (28). For electron microscopy the added advantages of 
osmic acid as a fixative are its ability to combine chemically with 
certain cellular constituents, thus increasing contrast, and its 
capacity to harden tissues to an extent which reduces the possi¬ 
bility of introduction of secondary artefact during the later proc¬ 
esses of dehydration, embedding and sectioning. Some of the dis¬ 
advantages of its use are its poor penetrating power (approximately 
0.5 mm/hr through parenchymatous tissues), its tendency to cause 
electron scattering either as a result of its precipitation within the 
nucleus in a reduced form or because of its chemical combination 
with nuclear components, and, after somewhat longer exposure to 
it and possibly for the same reasons, loss of detail in the cytoplasm 
also. Fixation by perfusion circumvents the 1st of these disad¬ 
vantages, and the addition of lanthanum nitrate to precipitate 
nucleic acids and of potassium dichromate to reduce the deposition 
of osmium in the tissues largely eliminates the latter 2. Thus in 
mice anesthetized by intraperitoneal injection of 1 cc of 0.2 per 
cent Nembutal containing 0.1 per cent sodium nitrite or sodium 

nitroprusside, perfusion with approximately 10 cc of cold 0.85 

per cent sodium chloride followed immediately by perfusion with a 
mixture of equal parts of 2 per cent osmic acid, 3 per cent potassium 
‘dichromate and 2 per cent lanthanum nitrate, followed by immer¬ 
sion of thin slices of the perfused tissue in the sarne mixture for 24 

hr resulted in satisfactory fixation and preservation of both cyto- 

1 .-M 1_^ 7 ^ /'Eicr IV in the 

dichromate at 37 C for 3 days, 


followed by treatment with 2 per cent osmic acid at room tempera- 
fure for 4-24 hr has given reasonably satisfactory results (see 

For^visualization of the Golgi substance, perfusion with modified 
Champy fluid (equal parts of 2 per cent osmic acid, 3 per cent 
potassium dichromate and 1 per cent chromic acid), immersion ot 
thin slices in the same mixture for 24 hr and vigorous washing in 
tap water for 24 hr and treatment with 2 per cent osmic acid at 37 C 
for 3 days has given excellent results (Fig. 5). 

It is hoped and fully expected that greater emphasis on and 
recognition of the primary importance of fixation of tissues for 

Fig. 3.—Part of a cell from tho convoluted jiortion of a proximal tubule of 
a mouse kidney fi.xed as the cell in Figure 1, showing detail of the brush border. 
.Approximately 11,000 X. 

electron microscopy will, in the future, bring forth more satis¬ 
factory methods of fixation than those described here. These 
methods have, in our hands, given more satisfactory results than 
methods previously described in the literature, e.g., perfusion with 
osmic acid alone (3), fixation by immersion in osmic acid (24) 
or Formalin (26) or fixation by immersion in a series of certain 
chemical agents (12, 13). 

There are several statements in the literature to the effect that 
removal of the embedding matrix before examination with the 
electron microscope, either following the double embedding method 
of 1 ease and Baker (1, 23) or the n-butyl methacrylate embedding 
method of Newman, Borysko and Swerdlow (18), results in the 

i.’rn d Spvpral cells of the epithelium of the colon of a mouse perfused and 

" Httle evk^nce of dis^ fixed by 

pe fSio^n^cK^y' fluid follow W 



introduction of relatively gross artefacts. In our experience, fixa¬ 
tives containing osmic acid or fixation methods involving secondary 
treatment with osmic acid not only reduce such artefacts but also 
largely eliminate the damage which frequently results during the 
polymerization of n-butyl methacrylate. 

Comment by Harrison Latta 

The fixation procedure Dr. Dalton describes certainly avoids many 
artefacts apt to appear with other methods. Further evaluation of the ef¬ 
fects of preparative procedures on cellular ultrastructure appears to re¬ 
quire study of less complex protein systems, correlating the results of elec¬ 
tron microscopy with those of other biophysical and chemical techniques. 
Indeed, most biologic problems which aim at interpreting cellular events 
in terms of molecular and macromolecular phenomena necessitate such a 
correlative approach. 

Additional discussions on thin sectioning procedures include refinements 
on a conventional microtome (15), a triple embedding procedure (22), a 
double sectioning procedure (27) and a study of the kidney using a rocking 
microtome (19). In the last study, the observation (24) of ridges on the 
external surface of the glomerular capillary basement membrane was con¬ 
firmed, and a honeycombed trellis on the interior of the membrane was 
associated with endothelial cells. Further experience with the glass knife 
(16) has sho^Ti that more consistent results are obtained if oil and dust are 
removed from the edge before use by dipping successively in an organic 
solvent, sulfuric acid-potassium dichromate cleaning solution, and dis¬ 
tilled water and drying without wiping. Excellent electron micrographs 
have been obtained recently from several tissues prepared with a specially 
built microtome (21). 

Various other important methods of preparing biologic specimens for 
electron microscopy are available or are in the process of development. 
Further references will be found in recent treatises on the practice of elec¬ 
tron microscopy (8, 29) or in the bibliographies on electron microsconv 
(4, 17). 

Comment by A. C. Corcoran 

The fields of tissue electron microscopy and ultramicropathology are 
just opening. The technique described is one way in which kidney tissue 
may be prepared for study under the electron microscope. The enormous 
vascularity of the kidney undoubtedly facilitates the procedure, so that ap¬ 
plications of this method m other organs may require adaptations of a 
technique which may, m its own right, illuminate some of the peculiarly 
dark and obscure corners of renal pathology. 

1 . 

2 . 


Baker, R. F., and Pease, D. C.: Improved sectioning techniques for elec- 
tron microscopy, J. Appl. Physics 20: 480, 1949 
Bretschneider, L. H. : A simple technique for the electron microscopy of 
ceii and taue aect.ona, Proc, Koninkl. Nederland. Akad. CteSap 





0 . 



8 . 

9 . 

10 . 

11 . 

12 . 








20 . 
21 . 
22 . 



Claude, A., and Fullam, E. F.: Preparation of sections of guinea pig liver 
for electron microscopy, J. Exper. Med. 83: 499-504, 1946. 

Cosslett, V. E.: Bibliography of Electron Microscopy (New York: Long¬ 
mans, Green & Co., 1951). 

Dalton, A. J.: Stnictural details of some of the epithelial cell types in the 
kidney of the mouse as revealed by the electron microscope, J. Nat. 
Cancer Inst. 11: 1163, 1951. 

Dalton, A. J.: Electron micrography of epithelial cells of the gastroin¬ 
testinal tract and pancreas. Am. J. Anat. 89: 109, 1951. 

Dalton, A. J., el al.: Finer structure of hepatic, intestinal and renal cells 
of the mouse as revealed by the electron microscope, J. Nat. Cancer 
Inst. 11: 439, 1950. 

Drummond, D. G. (ed.): The j>ractice of electron microscopy, J. Roy. 
Micro. Soc. 70: 1, 1950. 

Eden, M,; Pratt, A. W., and Kahler, H.: A microtome specimen holder 
advanced by thermal e.xpansion. Rev. Scient. Instruments 9: 802,1950. 

Geren, B. B., and McCulloch, D.: Development and use of the Minot 
rotary microtome for thin sectioning, E.xper. Cell Res. 2: 97, 1951. 

Gessler, A. E., and Fullam, E. F.: Sectioning for the electron microscope 
accomplished b}' the high speed microtome, Am. J. Anat. 78: 245, 


Gessler, A. E., el al.: Notes on the electron microscopy of tissue sections: 

I. Normal tissues. Cancer Res. 8: 534-548, 1948. 

Grey C. E., and Kelsch, J. J.: Use of the electron microscope and high¬ 
speed microtome in medicine, Exper. Med. & Surg. 6: 368—389, 1948. 
Hillier, J., and Gettner, M. E.: Some refinements of the rotary micro¬ 
tome modified for ultra-thin sectioning, J. Appl. Physics 21: 67 (abst.). 
Hillier, J., and Gettner, M. E.: Improved ultra-thin sectioning of tissue 
for electron microscopy, J. Appl. Physics 21: 889, 1950. 

Latta, H., and Hartmann, J. F.: Use of a glass edge in thin sectioning for 
electron microscopy, Proc. Soc, Exper. Biol. & Med. 74: 436, 195 . 
Marton, C., el al.: Bibliography of Eleclron Microscopy, Nat’l Bur oi 
Standards, circ. 502 (Washington, D. C.: U.S. Government Printing 

Office, 1950). , . . u 

Newman, S. B.; Borysko, E., and Swerdlow, M : U tra-microtomy by a 
new method, J. Res. Nat. Bureau Standards 43: 183, 1949. 

Oberling, C.; Gautier, A., and Berrihard, W.: 

glomerulaire vue au microscope electronique. Pi esse m6d. o. .. , ■ 

O’Brien, H. C., and McKinley, G M.: New microtome and sectioning 
method for electron microscopy, Science 98: 4oo, 194^. 

Palade, G. E.: A study of fixation for electron microscopy, J. Exper. 

Med' 95:285, 1952. . . • j u , 

Pease D. C.: Triple embedding for the ultrathin sectioning required by 

electron microscopy, Anat. Rec. 110: 531, 1951. niicros- 

Pease D C., and Baker, R. F.: Sectioning techniques for electron micros 

copy ushig a conventional microtome, Proc. So.^ E.xper. Biol. & Med. 

Peale ^D’c^nd Baker, R. F.: Electron micro.scopy of the kidney. Am 

J. Anat. 87: 349-389, 1950. ^ ^ .nierotome 

Richards A. G., Jr.; microscopy illustrated with sections 

1.01. Kled. 51: 148, 1942. 


26. Rozsa, G., and Wyckoff, R. W. G.: The electron microscopy of dividing 

cells, Biochim. et biophys. Acta 6: 334, 1950. 

27. Sjostrand, F.: Method for making ultra-thin sections for electron micros¬ 

copy at high resolution. Nature, London 158: 545, 1951. 

28. Strangeways, T. S. P., and Canti, R. G.: The living cell in vitro as shown 

by dark-ground illumination. Quart. J. Micro. Sc. 71: 1, 1928. 

29. Wyckoff, R. W. G.: Electron Microscopy (New York: Interscience Pub¬ 

lishers, Inc., 1949). 


Smith (14) has pointed out that, in view of the inherent technical 
difficulties, “it is surprising that five groups of investigators arrive 
at about the same figure for the filtration rate (0.6 to 0.76 cc. per 
minute per 100 grams BW).” Modes of collection of urine, of blood, 
of giving test substances, of establishing diuresis, all vary; varia¬ 
tions in diet and differences in strain must also be reckoned with. 
The very number of techniques that have been described attests 
the desire of investigators to have methods of renal study which 
they can apply in small animals. Consequently, 3 methods are 
described here in the hope that some synthesis of procedures may 
resolve the issues which remain. 

I. Measurement of Filtration Rate and TmpAH 

G. M. C. MASSON and A. C. CORCORAN, Cleveland Clinic Foundation 

Principle .—Estimation simultaneously of filtration rate from 
exogenous creatinine clearance (Ccr e^og) and of tubular secretory 
capacity in the rat: catheter collection of urine during osmotic 
diuresis, subcutaneous injection of test substances, estiination of 
plasma concentration by interpolation from 2 blood samples, hg t 
anesthesia before urine collection. 


a') Ar)varalus—il) Rat holder with sliding sleeve adjustable to 
antairof differenV^ (2) Pipets (^calibrated) for col 

blood; soft glass, 4 mm inside diameter. (3) Ureteral catheter^ 
no. 4 F, about 15 mm long. (4) Warming box set in the range 40 

Soluitions—{l) Heparin in 1 per cent NaCl (2 mg/cc). (2) 

PALr^annTtol-crUinin^: PAHNa 6 8.--f “n: 

4 g— all per 100 cc of sterile water. (3) Optional. Intiacai 


O lecnmqui. intrapentoneal injection ot 

0 3 f c oThetrr n solution and^lightly anesthetized with ether. 

9 F«Vh rat then receives a subcutaneous injection of 1.7 cc ot 


as 0. The bladder is catheterized and the rat is plaoeQ 




adjusted to restrain movement. Intracaine can be injected into the 
bladder to relieve discomfort as anesthesia wears off. 

3. At 40 min after 0 time, rat and holder are placed in the warmer 
for 3 min. The 1st blood (Bl) is collected onto a heparinized watch 
glass from the snipped end of the tail and drawn up into a pipet, 
the dry end of which is sealed in a flame. 

4. At 48 min after 0, the bladder is rinsed with 4 portions of 
saline (0.2-0.5 cc each); the last rinse is completed exactly at 50 
min and all rinses are discarded. 

5. From 50 to 58 min, urine is collected by drainage into a 
5 cc graduated cylinder. Rinsing is then repeated as before, ending 
at 60 min, and the rinses are added to the urine sample; urine + 
rinse is then made up to 100 cc with water. 

6. The rat is then warmed as before and blood (B2) collected as 
before, the collection ending at 65 min. 

7. For serial determinations in a group, each succeeding rat 
should begin 0 time at intervals of about 25 min. 

8. The pipets are then sealed at 1 end in a flame and the blood 
is centrifuged in them; they are then broken at the cell-plasma 
interface and 0.1 cc of clear plasma is secured for preparation of 
a CdS 04 -Na 0 H protein-free filtrate (1:100 dilution). PAH is 
determined in filtrate and diluted urine by the method of Goldring 
and Chasis and creatinine by an adaptation of the method of Folin 
and Wu. Mean plasma concentration during the clearance period 
is estimated by interpolation from a semilogarithmic plot of values 
of Bl and B2 against time. 

d) Calculations. These are made in the customary manner, using 
for Tm the value of 0.83 assigned to man. 

Results. Ccrexog averages 0.61 cc/min in Sprague-Dawley rats 
(2); a mean of 1.12 has been found in rats of the Wistar strain 
(12) and of 0.75 in the same strain on a creatine-creatinine-free 
diet. Mean TmpAH averages 0.29 mg/min in our Sprague-Dawleys 
0.47 m Wistars and 0.38 in Wistars on the creatine-creatinine-free 
diet (12). The values are stated in units/100 g body weight. Deter¬ 
minations repeated at intervals of weeks should not depart from 
the mean by more than 10 per cent (2). 

Evaluation.—There are several advantages. Accuracy of urine 
”i establishment of mannitol diuresis, cathe- 

interpolat on of plasma concentrations during the clearance period • 
artefacts due to anesthesia, trauma or blood loss are minimLd by 
making urine collections under physiologic conditions ^ 

Disadvantages are pointed out below. One which is not noted is 
the probable inaccuracy of the factor 0.83 for FW in the calculation 



of PAH; another is the tedious character of the experiment, which 
limits observations to 6 rats per morning or afternoon. 

Comment by Richard W. Lippman 

This method is similar to our “tail-cutting” procedure (9). The objec¬ 
tion to it is that tail-cutting causes circulatory responses which extend to 
the kidney and persist longer than is allowed for in steps 3 and 4. We have 
found subcutaneous absorption of test substances irregular in the rat, so 
that even 1 hr after injection one cannot be sure of a drop in serum con¬ 
centrations. We have also found that the high plasma concentrations nec¬ 
essary for determination of TmpAH depress inulin and creatinine clear¬ 
ances in the rat and for longer periods than in dog or man (10). 

The validity of creatinine clearance as a measure of glomerular filtration 
in the rat has also been questioned. Martel et al. (12) found that Ccr, de¬ 
termined as described here, fell slightly as plasma creatinine content in¬ 
creased, although the change is of doubtful significance. We had shown 
that Ccr rises conspicuously as the transition is made from measurement 
of endogenous to exogenous creatinine (8), and Martel et al. (12) found e.x- 
ogenous Ccr consistently greater than Cin. The ratio Ccr/Cin Martel ob¬ 
served was consistent and of the same order as that we have found. Pre¬ 
liminary experiments (our laboratory and Martel) indicate that this ratio 
not only is stable under a variety of conditions but is also not altered by 
the giving of phlorizin. We have therefore continued to use it as a measure 
of glomerular filtration rate in the rat, aware that it may be found not to 
yield absolute values thereof. 

Comment by S. E. Dicker 

Disadvantages of the method are that it allows for the use of female rats 
only; more seriously, it involves anesthesia (5, 6), and measurements of 
renal function may be altered by mannitol diuresis. 

Comment by Dr. Masson and Corcoran 

Martel and his colleagues (12) modified our method by decieasing con¬ 
centration of injected substances to about V 4 and by using only 1 end¬ 
point blood sample without, in their opinion and ours, substantially de¬ 
creasing the accuracy of the method. The use of the 1 end-point blood, 
previously recommended by Braun-Menendez and Chiodi (1) for deter¬ 
minations of Cm and Tmo, obviates Dr. Lippman’s objection to pre¬ 
liminary tail-cutting. His objection to the use of PAH, sm^ confirmed y 
Martel (12), stands, as do the disadvantages noted by Dr. Dicker. 

II. ‘‘Undisturbed” Method for Clearance 

RICHARD W. LIPPMAN, Cedars of Lebanon Hospital, Los Angeles 

Principle —Measurement of exogenous Ccr as an indicator of 
filtration i-nte; saline hydration, intravenous nijection ol test 


substances, interpolation of mean plasma concentrations from 
measurements in injected controls. 

PROCEDURE (1, 11) 

a) Apparatus. —(1) Metabolism cage and funnel. (2) Deflecting 
bulb and centrifuge tube (Fig. 1). 

h) Solutions.—{\) 0.85 per cent NaCl. (2) 0.85 per cent NaCU 
containing 5 mg creatinine/1.7 ml. This solution is prepared in 
sufficient volume to serve for control and test groups in any single 

c) Technique. —1. Rats of the same sex, 
uniform size and under stable dietary con¬ 
ditions are divided in 2 groups of 6. They 
are given 3 intraperitoneal injections of sa¬ 
line at 9:30a.m., 4:30 p.m. and 9:30 a.m. the 
following day. 

2. Five hr after the 3d intraperitoneal in¬ 
jection, the controls are given 1.7 ml of cre¬ 
atinine-saline by intravenous injection into 
the dorsal foot vein, using a tuberculin 
syringe and 25 gauge needle. 

3. Two min after completion of the intra¬ 
venous injection, each rat in this group is 
lightly anesthetized with ether, the abdomen 
opened and the rat killed by exsanguination 
from the abdominal aorta. The blood is col¬ 
lected lor creatinine determination. 

4. A similar intravenous injection from the 

same lot of creatinine-saline is given animals 
of the 2d group, which are then placed in 
cages foi collection of urine. In either group, 
leakage from the injection site is controlled 
by application of a small rubber tourniquet Do. 1. 

held m place with a Dieffeback clamp for as long as need be or 
the foot can be dipped in collodion. ’ 

group. AUhetre lime the ^ 

urine is evacuated by syringe and added ^ remaining 

collected. ® '‘""e previously 

6. Funnel and collection bulb are rin^pH u-uk n 
saline and the rinse fluids added to the urine IplciiZ 



d) Calculations .—Clearance is calculated from mean serum and 
measured urine creatinine concentrations. Mean serum concentra¬ 
tion is calculated on the assumption that concentration decreases 
exponentially after intrav^enous injection of the test substance. 

Let a = group mean serum creatinine concentration at 2 min; 
b = group mean serum creatinine concentration at 15 min; n = 
group mean serum creatinine concentration at the 15 min clearance 
period for any individual rat, and Uis = serum creatinine concen¬ 
tration of any individual rat at 15 min. Then, since the measured 
serum concentrations are 13 min apart, but the collection period 
lasts 15 min 

log n = log ni5 + 

7.5 (log a — log b) 

Thus, 6 individual clearances are obtained from each group of 12 

Results .—Under these conditions a mean creatinine clearance of 
1.36 ml/min/100 g body weight has been observed in rats of the 
Slonaker-Addis strain on creatine-creatinine-free diets. Sellers 
et al. (13) have obtained similar normal values by this method. 

III. Determination of Inulin Clearance 



S. E. Dicker, University College, London, has summarized for us 

the method which he has devised. 

PrmcfpZe.—Determination of inulin clearance as a measure ot 
glomerular filtration: establishment of water diuresis, subcutaneous 
injection of test substance, brief collection of spontaneously voided 

urine, and single (terminal) blood sampling. 

a) AvV<^ralus.-{\) Buret, 50 ml. (2) Soft rubber Jacques cathe¬ 
ter no. 41-3. (3) Glass funnel, 25 cm in diameter, fitted with feces 
separator which allows urine to collect into a finely graduate 

^^b)^Solutian.—Inulin, 5 per cent in 0.85 per 

c) Technique.-Rats of at least 150 g body weight are used 
(3 4) One is wrapped in a towel, a small gag placed in the inouth 
in’d L JaVcues c^^heter fitted at 1 end to the 
the stomach, and through it is given 5 ml. of Avater/100 g noay 
weight This is followed immediately by subcutaneous injec 
1 ml/100 g body weight of inulin solution. The bladder is emptied 
lo min lair by prodding, by suprapubic pressure or both. The 



animal is then put in the funnel and urine collected for 5-7 min; 
since it is in full diuresis, it will void spontaneously. (Postmortem 
controls have always shown the bladder quite empty.) The rat is 
immediately anesthetized and blood sampled either from the carot¬ 
id and jugular vessels or from the tip of the tail. 

Evaluation .—Advantages are: (1) the procedure is done by I 
worker only; (2) only 1 blood sample is required (50 minutes after 
subcutaneous injection of inulin) the concentration in plasma does 
not vary significantly during the short period of urine collection 
(6, 7), and (3) anesthesia at the end of the collection period cannot 
affect renal function as measured. 

Disadvantage is difficulty in administering water; Dicker writes; 
“after years of practice, I can manage 16 rats in less than 15 min.” 


■ S. J. Friedman, University of British Columbia, has summarized 
the procedure he uses in measurement of the plasma clearances of 
inulin and PAH (7). 



iO 100 220 260 

Weight in grama 

IiG. 1. Dosage table for PAH administration. 

Two solutions are prepared: (I) PAH, 12.5 mg/cc in 2 per cent 
sodium sulfate; (2) muhn, 2 per cent in 1 per cent NaCl. 

PAH solution IS injected subcutaneously in the lumbar region in 
an amount estimated from a dose table (Fig. I) to yield Sasma 
concentrations of 5-7 mg/IOO cc. Immediately thereafter 3^ 

Sly inloTw"'? " ‘"‘’■aP^ritoneally. The rat is placed 
M Th„ n n ® funnel over a beaker and secured there by a 
and tb IS emptied by suprapubic pressure 50 min later 

d the tail rinsed into funnel and beaker. Blood (0 75 cc) is then 
obta.ned by heart puncture, using a heparinized syri^Uan^ 



24 or 25 gauge needle; sampling should be completed by the 51st 
minute. The sample is prepared for analysis by centrifugation. 
The funnel is rinsed into the beaker and urine + washing made up 
with water to 100 ml. Any fecal admixture is removed by centrif¬ 
ugation of the diluted urine. 

Friedman notes: “The procedure is not too accurate, but is 
satisfactory in large groups of animals. With a team of 2, we can 
do 16 clearances in the morning and as many in the afternoon.” 
Values of Cpah and Cin thus obtained average respectively 4.1 
iind 0.65 cc/min/100 g body weight. 


In this method, inulin and Diodrast clearances and Tmo were 
determined after subcutaneous injection of solutions of test sub¬ 
stances and initiation ol water diuresis at 0 time by administration 
of water, equivalent to 5 per cent of body weight. The bladder is 
emptied by manual expression 1 hr later, the rat placed in a metab¬ 
olism cage and the bladder again emptied manually if the rat 
has not completely voided spontaneously at about 30 min. The 

blood sample is then secured. i n rn 

By this method CV and Cm average, respectively, 2.66 and 0.60 

cc/min/lOO g body weight, and Tmo (as iodine) 0.18 mg. Eac 
function shows some correlation with urine flow. 

General Comment 


The methods described here cair perhaps best be evaluated in 
terms of advantages and disadvantages, and it is obvious that the 
accuracy of each is more dependent on pooling of data 
than on extreme accuracy in individual animals. The obMoiis ais 
advantage of Lippman’s method is the need for killing not only test, 
but also control Limals; serial observations are impossible and the 

to tnc y u 1/-1 Kq imtpfl Ourinff the few minutes 

plasma concentration be ,mted^_ Diu^ 

alter in] t j j function, and, since 

occur volume effects on renal cirima ^ exponential dis- 

the curve of creatinine m P ,emoval and diffusion in tissue 

appearances, respectnely, y ,t 



seems by all odds the simplest in principle and as accurate any 
in practice. Friedman’s method may suffer from administration of 
PAH as a test substance, from the long periods of urine collection 
after injection and the choice of the end-point single blood sample 
as representative of this period. The method of Braun-Menendez 
and Chiodi has the advantage that urine is only collected after 
partial equilibration of blood levels and that the collection period 
is short and the blood drawn at the end of that time. However, 
again, the precision of an end-point blood sjimple for a 30 min 
clearance period can be questioned. 

Thus, the points to be resolved in the estimation of renal func¬ 
tion in rats seem to us to be as follows* 

1. C’an spontaneous voiding even during diuresis be relied on for 
complete collection of urine formed over short intervals? 

2. If, as seems evident, some diuresis must be established, at 
what level and by what means (water, saline, sulfate or mannitol)? 

3. What concentration of PAH will not depress glomerular fil¬ 
tration rate and what substance is to be used to measure true filtra¬ 
tion rate in the rat? 

4. If end-point blood must be used as the single sample, what can 
be done to stabilize blood concentration over the preceding collec¬ 
tion period? How long should that period be? What allowances 
must be made for abnormalities of renal function? 

1 . 

2 . 




0 . 


1 ). 

10 . 

11 . 

12 . 


14 . 


Braun-.Menendez, E., and Chiodi, H.: Rev. Soc. argent, biol. 22:314 

Corcoran, A. C., et al.\ .4m. J. Physiol. 1.54: 170, 1948. 

Dicker, S. E.: Science 108: 12, 1948. 

Dicker, S. E.: J. Physiol. 108: 197, 1949. 

Dicker, S. E., and Ileller, H.: J. Physiol. 103: 449, 1945. 

Dicker, S. E., and Heller, H.: Science 100: 127, 1947. 
triedman, S. J.; Policy, J. R., and Friedman, C.: .4m. J. Physiol loO* 
340, 1947. 

Lippman, R. W.: Am. J. Physiol. 151: 211, 1947. 

Lippman, R. W.: .4m. J. Physiol. 152: 27, 1948. 

Lippman, R. W.: .4m. J. Physiol. 155: 282, 1948. 

Lippman, R. W.; Ureen, H. J., and Oliver, J.: J. E.xper. Biol. 93- ‘i2b 
1951. ’ 

.Martel, I.; \tang, M., and Gengras, R.: Renal Function Studies in the 
^ Rat (Quebec: Presses Universitaires Laval, 1951) 

Sellers, A. I., et ai: .4m. J. Physiol. 166: 619, 1951. 

Smith, H W.: The Kidney: Structure and Function in Health and Disease 
(.New lork: Oxford University Press, 1950). 




The aspects of experimental renal pathology discussed here form 
a small segment of the whole topic. The selection is based on the 
premise, previously mentioned, that the rat has come to stay as a 
subject for the investigation of many problems in vascular and 
renal disease, although certainly not to the exclusion of other 
species. In particular aspects the vascular and renal, as well as 
social, responses of rats have perhaps more in common with those 
of human beings than do the reactions of other lower animals. 

In particular, the measurement of arterial pressure in such an 
animal as the dog is a well established procedure, whereas this 
determination is a vexed issue in the rat. The tail plethysmographic 
method of Harrison et al. (2) has been repeatedly modified; in our 
hands, in much the same form as was described by Kempf and 
Page (1), it has been on the whole satisfactory. However, it is still 
a commonplace for investigators who willingly accept their own 
measurements made with this apparatus to reject as faulty the 
measurements made in good faith in other laboratories. I he ne'w 
methods described here have such definite advantages of end¬ 
point that it seems likely they will supplant the older procedures. 

The establishment of hypertension in rats is a topic closely re¬ 
lated to the measurement of arterial pressure and, as will be seen, 

seems somehow usually to involve the kidney. , • • • 

Heymann’s technique of eliciting nephrotoxic nephritis in rats 
interestingly concludes a chapter which, left to his pleasure, the 
editor might have filled in still other ways of causing, relieving and 
studying renal disease. 


373, 1931). 



I. Measurement of Arterial Pressure in Rats 


MEYER FRIEDMAN and S. CHARLES FREED, Mount Zion Hospital, 

Sian Francisco 

The method of obtaining the systolic blood pressure of the rat is 
similar in principle to that employed in the indirect measurement of 
systolic blood pressure in man, namely, the detection of sound at 
the exact time that the arterial pressure of the caudal artery exceeds 
the pressure of the occluding cuff. This method was found to be 
accurate, simple and rapid. 


a) Apparatus .—-This consists essentially of a carbon type micro¬ 
phone 3 cm in diameter to the diaphragm of which is attached a 

L—Electric circuit of amplifier. F,, V^, type IT 4 vacuum tubes; 
i, otancor type A-4705 microphone transformer; J], J^, open circuit phone 
jacks; Ci, C 3 ,1 mfd by-pass condensers; C 2 , 0.1 mfd coupling condenser; C 4 , 
0.001 mica condenser; Ri, 1 megohm; R^, .500,000 ohms; R„ 1 megohm poten¬ 
tiometer; Rt, 100,000 ohms; R^, 100,000 ohms. 

thin copper trough 1.5 cm long and 0.6 cm in diameter. The micro¬ 
phone is connected to a specially designed, low frequency (100 
cps) sound amplifier (Fig. 1)* operated by direct current supplied 
by 3 dry cells. The amplifier affords a voltage gain of approxi¬ 
mately 3,000. The energy changes obtained by the amplifier may 
be detected by ordinary earphones, loud-speaker or oscilloscope. 

b) Techmque.~To take blood pressure readings, a rat is warmed 
5-10 mm m a wood box, the temperature of which is maintained at 
ap proxima tely 37 C. The tail of the rat is led through a 10 mm 

be^furnSed b^th^uS^ the electrical circuit wUl 

obtained froin^Charles Calvert of thp microphomc manometer itself may be 

378-28thSt..San Franc^co Industrial Electronics Comp^, 

•ji... oau r rancisco, i_.ant. 1 he cost is approximately $200. 



pressure cuff and the segment of the tail immediately distal to the 
cuff is placed in the trough and fixed by a strip of adhesive tape. 
The pressure of the cuff is raised with a rubber bulb to a point 
above the expected blood pressure, then slowly released. At the 
point at which the cuff pressure becomes less than the caudal ar¬ 
terial pressure, one hears immediately in the earphones a rhythmic 
succession of sounds reflecting the transmitted pulsatile variations 
of the caudal artery. If the amplifier is connected to an oscilloscope, 
one sees at this same critical point an intense and abrupt change in 
contour and frequency from the preceding waves. It should be 
stresse<l that the regularity and actual rate of these sounds do not 
allow their confusion with any possible extraneous movements of 
either the body or the tail of the rat. As the cuff is deflated further, 
the sounds attain their greatest intensity at the expected diastolic 
level, although this increase in intensity is too gradual to allow 
precise determination of diastolic pressure. The sounds continue 
even when the cuff is completely deflated. 

Preliminary venous drainage of the tail is not necessary as the 
method depends not on gradual increase in tail volume but on 
the almost instantaneously transmitted pulsations which occur 
when the cuff pressure is overcome. 

Comment by IP. Glen Moss 

My impression is that the combined “audiovisual” procedure is the best 
indirect method available for determining blood pressure in the rat. Dif¬ 
ficulties may be encountered from respiratory and other movements 
which are transmitted through the rat’s tail and the framework of the 
holder and chamber to the microphone. We have overcome most of 
bv seating the microphone over rubber and applying a rubber-lined clamp 
laterally to the tail just before it enters the cuff. In some rats with e.xperi- 
mental renal hypertension, we have heard pulsations when the cuff pies- 
sure was well above systolic blood pressure. This seems to be due to a par¬ 
tial shielding of the “constricted” caudal arterj'^ by the small processes on 
the caudal vertebrae; however, this is only a theory. If the tails of such 
animals are warmed by washing in water at 45-50 C and the measurement 
repeated, a satisfactory reading can usually be made, presumably because 
the caudal artery has vasodilated. The procedure is readdy taught, and 
by using 3 rat holders in rotation, a rate of 20-40 measurements/hr can be 

In rats with experimental hypertension in which sounds can be heard at 
all cuff pressures, the same difficulty is found in attempting to detennme 
systolic pressure by the plethy.sinographic technique. Under anesthesia, 
he blood pressure Lorded by direct artery puncture ,s w. hm the rarrge 
of the manometer used in the cuff method. It is because of this that I be¬ 
lieve the caudal artery in such rats is “protected” from the pressuie ap¬ 
plied through the cuff. Therefore in such animals it would seem that the 



onlv accurate means of measuring systolic pressure is by cannulation or ai- 
toy puncture. I have not investigated this difficulty m nrmu™g the blood 
pressure of hypertensive rats any more thoroughly than stated here. 

recording of indirect blood pressure in the un- 
^ anesthetized rat 

Fl^KDERICK OLMSTED, Cleveland Clinic Foundation 

Graphic registration of foot pulse in the rat, with simultaneous 
record of air pressure in an occluding cuff proximal to the point oi 
pulse pick-up, offers a method of blood pressure determination m 
this animal free from subjective error, with the advantage of a per¬ 
manent recording for each analysis. Ihe method has been used, 
in addition to normal and hypertensive white rats, in Kangaroo 
rats, parabiotic rats and in rabbits; with minor modification, it 
seems applicable to most animals. In the rat, foot arteries must be 
at least normally dilated to insure a clear pulse recording and 
adequate compression of the pulse. Measurements are therefore 
made at an ambient temperature of 30 C. To avoid the pressor 
effect of abrupt heating (4), the animals are placed in this environ¬ 
ment several hours before measurements are made. 


a) Apparatus .—The rat holder, foot platform and electrical 
pick-up are sliown in Figure i. The center section of the holder is 
brass tubing; this varies from 1 hT to 2\/2 in. in diameter to accom¬ 
modate animals of different sizes. The front section is a heavy 
bronze mesh cone soldered to a brass cylinder which slides snugly 
in the center section. The rear part is consti’ucted of inner tube 
rubber, with the leg hole so positioned that the rat’s hip is locked 
in place by gentle squeezing of the rubber sides at the location of the 
clip. The unl)onded strain gauge* and foot platform are mounted 
at an angle of 45 degrees on a heavy metal slab that can be ma¬ 
neuvered horizontally to accommodate the foot. The oilcloth strap 
is Vs iu. wide and acts both as foot holder and pulse pick-up. 

The electrical circuit consists of slightly modified standard 
equipment (hig. 2). With minor alterations, the stage following the 
Grass preamplifier consists of the Inrush a-c amplifier for use with 
the Brush Penmotor. The arrangement gives a 3 cm pen deflection 
for a 50 wave applied at the output terminals of the strain 
gauge. However, any amplifier with a voltage gain of 6 X 10® 
with response attenuated 95 per cent below 5 and above 35 cps, 

les* #G1-1C-250, made })y Statham Co., 12401 W. Olympic Blvd., Lo.s Ange- 


and impervious to surge currents is satisfactory if the output im¬ 
pedance matches that of the ink oscillograph used. It is advisable 
to locate the sensing element physically so that microphonic inter¬ 
ference is avoided. 

The recording unit is a Brush Penmotor mounted at one side of a 
2-(‘hannel chassis. The paper width (double channel size) is 8.5 

cloth strap 



Li^ht clip 

Plastic air 



1 —.Animal holder and strain gauge mounting with animal in place (4). 

cm to allow easy determination of cuff pressure. The receptor 
selsyn fixed in the unoccupied Penmotor space^ (GE #2JIFI, 110 
V 400 cps, run on 25 v, 60 cps) actuates the cuff pressure pen normal 
to the direction of paper travel along a guide wire by a continuous 

thread belt driven by a drum on the motor shaft 

The driver selsyn is activated by a 4 in. sylphon bellotys in the 
pressure line opposed by a helical spring A simpler hut less con¬ 
venient arrangement would be to drive the pen directly from an 

t Metro Industries. 29 28 41st .\ve., Long Island City 1. N. Y. 


air bellows mounted on an oscillograph chassis. The air pressure 
pen is calibrated stepwise with a mercury manometer, its range 
being from zero to 250 mm Hg. 

The cuff described by Kersten and co-workers (3) and modified 
by a manufacturer^ is satisfactory for use with the apparatus. 
A simpler and more easily applied, but probably less accurate. 

6F5 6SF5 6Sf5 6V6 

Stalh am 


Fig. 2.—Electrical circuit (4) as described in text. The band pass type of circuit 

was found impractical. 

Ih Is U„ Ts h \ f ■“■Penrose tubing (4). Its effective 

anda smil here dam, 

To apnTtL " rff^ s^f ^ polyvinyl tube supplies air pressure! 
f 4 - ^ Single loop is made around the rat’s Ipp* ond 



lines reduced by a commercial valvet to 250 mm. Hg (about 5 
Ib/sq in.). Air is admitted to the cuff and recording bellows by a 
special valve§ (Fig. 3) whose action is such that pushing the 
control disk down inflates the cuff rapidly, and releasing it starts a 
gradual leak from the system, finely regulatable by rotation of the 
control disk. 

h) Technique .—Although the foot pulse method seems at first 
to involve more manipulation of the animal than some other meth- 


4 —Sphygmograms illustrating systolic points at downward 
^ nnintiue: arrows (4). 

ocls or tl.eir van.tions no. in KhAnS 

" Tht SSC- seenrs a Mt 

t Model #40, Harris Calorific Conipany^CleveU^ ^ 

\ Micro Metric Instruments Co., 28.n t.. / .an ^ 



too small for the rat to be tested) and the front cone pushed in until 
the animal is firmly but not uncomfortably compressed. The left 
leg is brought through the side hole, and the hip locked outside, 
as in Figure 1. The rear part is then refastened to secure this 
locking position. The holder is attached to a large buret clamp 
on a fixed vertical rod and the gauge platform moved to support 
the foot without altering its natural position. The pick-up strap 
is next placed over the foot at about the end of the metatarsal 
bones. \\Ten the adjustable block is clamped, the strap should not 
touch the sides of the slot in the platform. Tension should produce 
a snug fit around the foot, but not enough to distort its shape, 
agitate the rat or occlude venous return. 

The recorder is turned on for inspection of the pulse tracing. 
If this is satisfactory, the machine is halted while the cuff is ap¬ 
plied. Readjustment of the strap may then be necessary. To 
obtain a sphygmogram, record 1 or 2 in. of clear pulse beat and 
depress the valve disk until the cuff pressure rises to its limit or 
until the pulse tracing disappears, leaving only artefacts of respira¬ 
tion and animal movement. Then rotate the valve control disk 
counterclockwise until the cuff pressure starts an appropriate rate 
of fall. 

The relative systolic point is taken at the 1st reappearance of the 
pulse wave, which forms the apex of a cone of rhythmic pen fluctua¬ 
tions. Readings may be recorded as rapidly as this cycle of opera¬ 
tions can be repeated, with the cuff pressure allowed to fall to 
zero between measurements (Fig. 4). 

Note.— This section was reviewed by Simon Rodbard. 


1. Friedman, 'SI., and Freed, C. F.; Microphone manometer for the indirect 

determination of systolic blood pressure in the rat, Proc. Soc. E.xner 
^ Biol. & Med. 70: 670, 1949; and this volume, j). 251. 

2. Kempf, G. F., and Page, I. H.: Production of experimental hypertension 

Sin £'' 27 ! pressure in rats, J. Lab. & 

3. Kersten, H., et al: New method for the indirect measurement of blood pres¬ 

sure m the rat, J. Lab. & Clin. Med. 32: 1090 1947 

4 . Olmsted. F.; Corcoran A. C., and Page. I. H.:’Blood pressure in the r.nan- 

esthetized rat, Circulation 3: 722, 1951. 

i indirect determination of blood pressure in rats- 

PhyJ^^oTlS: ur/m and width of cuff. Am. J.’ 

6. Williams, J. R., Jr.; Harrison, T. R., and Grollman, A.: Simple method for 
InveTclg'sTrim raW Clin 



II. Establishment of Hypertension in Rats 

Experimental hypertension of varying severity and seemingly of 
distinct mechanisms can be established in rats by (1) initiation of 
a persistent change in renal circulation, as by compression of the 
renal artery or of the renal parenchyma (renal hypertension), (2) 
administration of substances possessing hormonal activity (hor¬ 
monal hypertension), and (3) provision of excess sodium (salt 
hypertension). In each instance (described in the following articles), 
the ultimate mechanism of hypertension is obscure; however, each 
of these forms of hypertension has some renal connotation. 


G. M. C. MASSON and A. C. CORCORAN, Cleveland Clinic Foundation 

Principle .—Placing the kidney in a rigid shell results in a change 
in renal circulation of the blood and tissue fluid which is appar¬ 
ently distinct from ischemia as such and which causes persistent 

elevation of arterial pressure (2). 

The foreign body reaction of the rat to perirenal application of 
silk or cellophane is much less severe than in the dog, and the 
fibrocollagenous hull that forms is consequently thinner. The 
tension with which the kidney is compressed by the wrapping at the 
time of its application is therefore a determinant of the succps of 
the procedure. This precise tension must be learned by experience. 


1 Rats weighing 70-100 g are selected because in the period of 
bodv growth, growth of the kidney tends to increase the com¬ 
pression exerted at operation. Males are preferred, because of the 
ovarian fat pads around the caudal pole of the kidneys of females. 

2. A jar is prepared for administration of ether anesthesia. 

3 Surgical scissors, large forceps, trachoma forceps, a pro e, 

suture needles and cotton thread are made ready. 

4 A set of rectangular pieces of closely woven silk of sizes varying 
around 35 X 25 mm with running stitches at the edge is made up^ 
From each silk piece the 2 ends of the running purse-string stitch 
are left free for tightening at the time of application 

5. The rat is anesthetized and placed on its side. The liimbai 

SktTnd ‘ 

forceps inserted and used in exteriorizing the kidney. The kidncj is 



freed of its surrounding fat by the probe without damaging its 
capsule. A silk square which will approximately fit the kidney is 
then applied to it, and the ends of the running stitch around the 
edge are pulled and tied. The wrapping should cover the whole 
kidney. It should not extend over the renal pedicle, nor should the 
pedicle be involved in the tie. The kidney is then placed back in the 
abdominal cavity and the muscle and skin wounds-are closed with 
a single suture. 

Results.—The severity and incidence of the hypertension which 
follows perirenal application of silk vary. If only 1 kidney is 
wrapped and the other is intact, hypertension results in about 20 
per cent of the animals, of which only occasional ones will show 
arterial pressures greater than 200 mm Hg. If the contralateral 
kidney is removed at the time of kidney wrapping, initial mortality 
from uremia will be large, but animals that survive the 1st post¬ 
operative week usually have severe hypertension which, in a few 
weeks, enters a malignant phase and causes death. If both kidneys 
are wrapped, hypertension of long duration and moderate severity 
appears in about 80 per cent, so this is the most generally satis¬ 
factory method. Periodic checks of pressure should be made during 
the first 3 weeks after operation in the last class of animals. Hyper¬ 
tension should have been sustained for 3-4 weeks before it can be 
considered stable. 

Other methods. —Abrams and Sobin have established renal 
parenchymal compression with latex capsules (1) and Rau by the 
application of a plastic in a solvent (3). We have no experience with 
either method; on the face of it, the method here described seems 
simpler and, with practice, as effective as any. 

Note. —This section was reviewed by Simon Rodbard. 


1 . Abrams, M., and Sol)in, S.: Proc. Soc. Exper. Biol. & Med 64-412 1947 

2. Page, I. H.: J.A.M.A. 113: 2046, 1939. 

3. Rjiu, G. C.: Science 111: 229,1950. 



DON.\LD E. GREGG, Army Medical Service GraducUe School 

A small, adjustable silver clamp modeled after the larger Gold- 
blatt clamp (1) has been designed and applied to the renal arteries 
ot the rat to induce hypertension (Fig. 1). The clamp (4) is 1.7 
mm high, 1.65 mm long and 1.5 mm wide. 



For application of the clamp, the renal arteries are exposed 
with aseptic precautions through lumbar incisions and under pento¬ 
barbital sodium anesthesia (4 mg/100 g intraperitoneally). A 
clamp is generally applied at 1 operation to each renal artery after 
it has been cleared for a distance of 3-5 mm, about 1 mm from the 
aorta. The removable plate of the clamp is taken out and held with 
a pointed hemostatic forceps. With a curved, blunt-pointed iris 
forceps modified to hold the clamp, the inverted clamp is slipped 


0 — i I I 

L_j _I _^_I_^_L_ 

0 12 3 


P’lG. 1.—Dinieti'ic drawing’; to scale of tlie adjustable renal clartij). 

under and up around the renal artery. The removable plate is 
seated and aligned. The clamp is then rotated through 180 to 
bring the screw into view. The screw is then turned down to just 
give complete occlusion and then backed off about F /2 turns with 

a small jeweler’s screw driver. , , , 1 . 

Results —After application of the clamps blood pressure valuer, 
as determined in the hindleg of the unanesthetized rat by a moc ih- 
cation (3) of the method of Kersten etal. (2), increased from contro 
levels of 100-125 mm Hg to at least 160 mm Hg within a period 
varving from as short as 8 days to as long as 21 days postopera- 
tively (3). The pressures remained elevated (160 up to a maximu 1 


of 220 mm Hg) for months. With careful technique V 4 ov more of 
the rats in different series developed hypertension of this degree 
and duration. As indicated macroscopically and microscopically, 
the kidneys of the rats were generally normal, comparing favorably 

A\ith those of unoperated controls. 

Since this method gives a high percentage of hypertensive rats, 
it is particularly useful for study of small p-oups of rats. Although 
the clamps can be used repeatedly on different rats, expense of 
construction precludes their use on large groups of animals. 


1. Goldblatt, n., et nl.: Studie.s in e.xperimental iiypertension: Production of 

persistent elevation of systolic blood pressure by means of renal ischemia, 
J. E.xper. Med. 59: 347, 1934. 

2. Kersten, H., et al. \ new method for the indirect measurement of blood 

pressure in the rat, J. Lab. & Clin. Med. 32: 1090, 1947. 

3. Munnel, E. R., and Gregg, D. E.: The production of chronic .systemic hy¬ 

pertension in the rat with a small adjustable renal artery clamp, J. Lab. 
A: Clin. Med. 36: 660, 1950. 

4. Munnell, E. R., and Gregg, D. E.; A .small adjustable renal arterj^ clamp 

for production of chronic hypertension in the rat, Proc. Soc. h]xper. Biol. 
A:-Med. 73:645, 1950. 


G. M. C. M.\SSON and .4. C. CORCORAN, Cleveland Clinic Foundation 

Among substances having some of the actions of hormones, 11- 
desoxycorticosterone (DC) or its acetate (DCA), 17-hydrox.y- 
desoxycorticosterone (compound S), diethylstilbestrol, thyroxin, 
crude anterior pituitary powder (APP, lyophilized anterior pitui¬ 
tary or LAP) and growth (somatotrophic) hormone have been 
reported to induce hypertension which, in certain instances, is 
a.ssociated with diffuse vascular disease and nephrosclerosis. 

Reports on the hypertensive properties of cortisone and com¬ 
pound F in the rat are relatively indefinite, largely because the 
experiments are complicated by cachexia and infection. 

Our own experience is limited to the use of DCA (2, 5), compound 
S (3), LAP (2), growth hormone (4), cortisone and compound F 
(1). In our hands, compound S does not induce hypertension- 
the hypertension caused by LAP and by growth hormone is vari¬ 
able both m degree and in incidence. Since the hypertension elicited 
by administration of DCA is relatively predictable in its course 
here occurrence, it is selected for description 

Principle.—The hypertension which results from administration 



of DCA seems to result in some way from the sodium-retaining 
effects of this steroid. 


1. Male and female rats weighing 60-80 g are selected and the 
left kidney is removed, since uninephrectomy sensitizes to the 
action of DCA. 

2 . The diet given is composed of commercial pellets; a solution of 
1 per cent NaCl is substituted for drinking water. 

3. The effectiveness of DCA depends on slow and constant ab¬ 
sorption of the steroid. This is most economically secured by 
subcutaneous implants or by repeated injection of the steroid 
acetate as a water suspension. The more convenient method is 
introduction of 2 pellets of 35 mg each of DCA powder prepared as 
flat disks 6 mm in diameter. These are prepared in a suitable die 
and press. The press consists of a hydraulic auto jack firmly at¬ 
tached to a heavy, base and supplied with a gauge connected with 
the oil reservoir. Depending on the consistency of the DCA powder 
supplied, the necessary pressure varies between 1,000 and 2,000 
Ib/sq in. 

4. The pellets are implanted at the time of uninephrectomy; 
blunt forceps are pushed in the subcutaneous tissue from the skin 
incision forward to each scapular region and a pellet is then in¬ 
serted at the end of each of these tracks. 

When water suspensions are preferred, the steroid is dissoh'ed 
in a minimal volume of hot alcohol and precipitated out bj' addi¬ 
tion of excess 1 per cent NaCl in water. Crystals which may form 
can be broken down by shaking with glass beads, a diop of deter¬ 
gent solution is added\,o maintain the suspension. The daily dose 
of 2.5 mg DCA is then given subcutaneously in 2 doses. 

Results .—Arterial pressure will be found to be increased by about 
30 mm Hg in 10-12 days, at which also the animals manifest 
definite polyuria. Hypertension, with pressures greater than 150 
mm Hg, will be present in all the animals by the 20th day. In 
some animals the disease pursues an acute course; m others the 
onset of manifestations of severe cardiovascular disease is delayed 
8-12 weeks. These manifestations include arterial pressure greater 
than 200 mm Hg, tics and convulsive phenomena and evidence o 
mesenteric arteritis and of advanced nephrosclerosis. 

—This section was reviewed by Simon Rodbard. 

1 . 

2 . 


r> Af n QTifl Coroorin A C.: Unpublislied observations, 
isson, G. M. C., and L^orcoran,1 ,«r Clin Med. 

isson; G. M. C.; Corcoran, A. C., and Page, I. H.. J. Lab. & t.lin. leo 
H: 1416, 1949. 



3. Masson, (5. M. C.; (Corcoran, A. C., and I’af^o, 1. II. : Endocrinology 46: 441, 


4. Selye, H.: Brit. M.J. 1:263, Feb. 10, 1951. 

5 Selye, H.; Hall, C. E., and Howley, E. M.: Canad. M.A. J. 49: 88, 1943. 


G. M. C. MASSON and A. C. CORCORAN, Cleveland Clinic Foundation 

Administration of excess sodium in the form of sodium chloride 
induces hypertension in rats (1, 2). We consider DC A hypertension 
a special and more severe form of salt hypertension, aggravated by 
the sodium-retaining properties of the steroid. However, since the 
procedure and the results differ significantly, the NaCl method is 
described here. 

Principle .—Sodium excess induces hypertension in rats sensi¬ 
tized by uninephrectomy. 


1 . Rats weighing 200-300 g are selected and subjected to left 
uninephrectomy. Large animals such as these object less strongly 
to substitution of hypertonic NaCl for drinking water and have 
less diarrhea and loss of weight during its administration than 
smaller animals. 

2. It is sometimes advantageous to start by substitution of 1 
per cent NaCl for water; however, it is possible to proceed without 
excessive mortality directly by administration of 1.75 per cent 
NaCl and, after 3 or 4 weeks, substitution for this of a 2 per cent 

Results .—Mean blood pressures of about 160 mm Hg (range 
140-200) will be found in the survivors after 2 months on a hyper¬ 
tonic regime, during at least 4 weeks of which 2 per cent NaCl 
has been given. This hypertension persists as long as treatment with 
hypertonic saline is continued. In our experience, it differs notably 
from DCA hypertension in that mesenteric vascular lesions do not 
occur and nephrosclerotic activity is minimal or absent. Whether 
or not this speaks for a pathogenic difference rather than a mere 
difference in severity is unknown. 

Note.—T his section was reviewed by Simon Rodbard. 


’■ '’Ssjosi - ’‘"‘I Lab. & Clin. Me<l. 




III. Nephrotic Syndrome Induced by Injection of 
Anti-Kidney Serum 

WALTER HEYMANN, Western Reserve University 

For the experimental production of renal disease in laboratory 
animals heteronephrotoxic antibodies were used for the first time 
by Lindemann (9) in 1900. Various workers used the idea sub¬ 
sequently. A thorough study of this type of renal disorder was 
made by Masugi (10) in 1923 in rabbits and by Smadel, Swift and 
Farr (13) 8 years later in rats. It was felt, however, that the value 
of these studies was limited because the experimentally produced 
disease did not seem to correspond clearly to any of the various 
forms of nephritis observed in man. In 1948, and in more extensive 
studies in 1951, Heymann and Lund (2, 3) showed that the disease 
produced in rats by the injection of anti-rat-kidney sera obtained 
from rabbits simulates the nephrotic syndrome—nephrosis with 
and without nephritis (1)—as it is seen in infants, children and, less 
frequently, in adults. 

The disorder is characterized in rats given injections of rabbit 
serum by immediate onset of massive proteinuria without gross 
hematuria, with edema, ascites, hypoproteinemia and hyperlipe¬ 
mia. Am initial azotemia subsides in the first 2—4 weeks. Spontane¬ 
ous cures do occur, but more freciuently proteinuria continues 
years without further injections. Hypertension, azotemia and 
creatinine retention may develop during the later stages of the dis¬ 
ease and the glomeruli then show thickened basement membranes, 
adhesions and possibly fibrosis and hyalinization. Many of the ani¬ 
mals, however, remain free trom clinical evidence of piogressive 
glomerular involvement and histologically the glomeruli are of 
normal appearance. One and the same experimental procedure thus 
may produce nephrosis in its so-called “pure” or “mixed” form. 

Heymann showed that the antigenic property of renal tissue is 
localized in the renal cortex (5). Solomon and colleagues (14) 
found soon thereafter that suspensions of glomeruli were antigenic, 
whereas suspensions of tubular proteins were inactive Krakower 
and Greenspon (8) have separated the basement membrane Irom 
the glomerular tufts and reported suspensions of basement mem¬ 
branes to be 50 times as antigenic. 


a) Preparation of antisera. —Antisera agmnst rat kicneys aie 
nrepared by injecting s\ispensions of perfused rat kidneys into la )- 
bits For pLfiising the kidneys 0.9 per cent ^aCI solution is ri^ 
into the thoracic aorta at a hydrostatic pressure of 1 m until 


kidneys are grossly free from blood. When the perfusion fails, the 
animals are discarded. The kidneys are then removed, weighed and 
0.9 per cent NaCl solution is added to make a 50 per cent suspen¬ 
sion. A fine suspension is obtained by placing the material in a War¬ 
ing Blendor for 5 min. After toluene and penicillin, 1,000 units/cc, 
are added the suspension is placed in an incubator at 37 C for 24 
hr. The material is cultured and preparations giving any bacterial 
growth are not used. The suspensions are stored frozen. Rabbits 
weighing about 2,500 g are given intraperitoneal injections of 10 cc 
of a 10 per cent, 15 per cent and 20 per cent suspension every 
other day for the 1st week. Twenty per cent suspensions are used 
for the 2d and 3d weeks. Six days after the last injection 20-30 cc 
of blood is obtained by cardiac puncture. This treatment may be 
repeated after 2 weeks’ rest 3-5 times in the same rabbit. The pro¬ 
cedure almost regularly yields highly nephrotoxic sera. 

TABLE 1.— Amount of Nephrotoxic Serum Injected into R.\ts, Gauged 

According to Kidney Weight 

Wt. op Rat, q 

Wt. of Kidneys, g 

Amt of Sbru 


































The rabbit serum is heated at 56 C for V 2 hr to diminish its primary 
toxicity. It is then stored frozen until used. Biologic assay is thus 

far the only means of determining the amount of nephrotoxin in an 

b) Analysis in rats. It has been shown that after intravenous 
injection, renal antibodies are rapidly and selectively absorbed in 
e kidneys (11). It thus is advisable to gauge the amount of 
nephrotic seruin to be injected into a rat according to kidney weight 
rather body weight. A series of kidney weights of healthy rats 
weighing 30-50 g is obtained from Table 1 and plotted. The weight 

ald“l in any rat to given injections, 

nendint nn fractions or multiples thereof, de¬ 

pending on the nephrotoxic activity of the serum, per 1 g rena tis- 

e:tioVr:l!hn?3^ "'f ^ b/intmUZs t 

jeciion on each of 3 consecutive days. 

Extensive experiences have been obtained with rats of the Long- 



Evans and Whelan strain. Their susceptibility to the disease does 
not differ. Rats of the Whelan strain have, however, a greater tend¬ 
ency to a progression of the glomerular lesion to fibrosis and 
atrophy (6). The younger rats have a greater tendency to anasarca. 
They may be given injections at the age of 4 weeks, weight approx¬ 
imately 50-60 g, 1 week after having been weaned. Quantitative 
determination of the 24 hr protein excretion is recommended. The 
physiologic proteinuria of the rat that varies with the strain of rats 
used and a dependence of the excreted urinary proteins on the 
urine volume make the quantitative estimation of qualitative tests 
for urinary proteins unreliable. For quantitative estimation of uri¬ 
nary protein excretion the method of Shevky and Stafford (12) is 
practical and reliable. For collection of urine the rats are placed in 
metabolism cages. The receptacle is to be covered with a rubber 
ring to prevent evaporation. For preservation of formed elements it 
should contain 2 drops of a 6 per cent Formalin solution. Methods 
for taking the blood pressure of rats are described elsewhere. We 
have been using the procedure of Kersten, et al. (7). Age-condi¬ 
tioned differences in the systolic blood pressure of the rat have to 
be taken into account and normal values, obtained with that 
method, have been established for rats weighing 50-400 g (4). 
For rats weighing 50—100 g, not more than 90 mm Hg should be 
considered normal; for rats weighing 100-350 g, 120 mm Hg, and 
for rats weighing over 350 g, 140 mm Hg. 

Comment by Ahbic I. Knowlton and Beattice C. Scegal 

In our experience the course of the disease and type of renal lesion pro¬ 
duced by rabbit anti-rat kidney serum parallels closely that reported by 
Dr Heymann, although early marked fluid retention has been observed 
in relatively few of our rats so treated. The technique employed in produc¬ 
ing and administering the serum in our laboratory is essentially similar o 
Dr. Heymann’s, but a few minor differences should be mentioned as these 
may possibly account for differences in our results. 

1. After mincing the rat kidneys in a Waring Bkndor, we store the ma¬ 
terial immediately in the frozen state, 

C which Dr. Heymann has employed. It is possible that this pei lod of me 
bation produces certain changes in the antigenic propert'®* ‘’' 5 ® 3 ^ 

2 In our laboratory the immunized rabbits are not bled until after 3 
courses of injections, and subsequently are bled, following provocative im- 
ZnStiX for periods up to a year or more. Dr. Heymann, on the other 

hand has bled rabbits after each course of injections. 

rChTe pooled the sera obtained from several individual rabbits 
to obWn larre TOlumes of a uniform character. This may be o real impor- 
tonce in eStag the lower incidence of edema seen, since a single edema- 
^rrulT—ay well have been ‘'diluted” by such jKiolmg. Dr, Hey¬ 
mann reports that he has tested e.ach serum separately. 

nephrotic syndrome from ANTI-KIDNEY SERUM 267 

4 . We have used, on the whole, older animals than Dr. Heymann and 

relatively smaller amounts of anti-kidney sei um. ^ 

It should be mentioned that we have seen the acute edema in ^prague- 
Dawley as well as the Long-Evans strain of rats. e have determmed blood 
pressure during the course of the disease. In agreement with Dr. Heymann, 
we have not seen hypertension during the early phase, even m severely 
edematous animals, but have observed it in the later phases (3 12 mont s). 
In severely edematous rats we have obtained markeiUy elevated nonpro¬ 
tein nitrogen values within a few days of serum administration. 

Comment by Dr. Heymann 

Although the factors indicated in comments (2) and (3) may explain 
some of the differences in the effects observed after injection of nephrotoxic 
sera, I believe that those indicated in comments (1) and (4) are perhaps 
more important determinants. 


1. Heymann, W., and Alperin, L. J.: Am. Pract. 3: 680, 1949. 

2. Heymann, W., and Lund, H. Z.; Science 108; 448, 1948. 

3. Heymann, W., and Lund, H. Z.: Pediatrics 7: 691, 1951. 

4. Heymann, W., and Salehar, M.; Proc. Soc. Exper. Biol. & Med. 72: 191, 


5. Heymann, W.; Gilkey, C., and Salehar, M.; Proc. Soc. Exper. Biol. & Med. 


6. Heymann, W.; Lund, H. Z., and Hackel, D.; J. Lab. & Clin. Med. 39: 218, 


7. Kersten, H., etal.: J. Lab. & Clin. Med. 32: 1090, 1947. 

8. Krakower, C. A., and Greenspon, S. A.; Arch. Path. 51: 629, 1951. 

9. Lindemann, W.; Ann. Inst. Pasteur 14; 49, 1900. 

10. Masugi, M. Beitr. path. Anat. u. allg. Path. 91: 82 and 92: 429, 1933. 

11. Pres.sman, D.; Hill, R. F., and Foota, R. W.; Science 109; 65, 1949. 

12. Shevky, M. C., and Stafford, D. D.; Arch. Int. Med. 32: 222, 1923. 

13. Smadel, J. E.; Swift, H. F., and Farr, L. E.; J. Exper. Med. 64: 921, 1936; 

65:527, 557 and 541, 1937; 74; 345, 1941. 

14. Solomon, D. H., etal.: J. Jlxper. Med. 90:267, 1949. 


Immunochemical Methods 
for Determining 
Homogeneity of Proteins 
and Polysaccharides 



There are 2 general methods for studying the homogeneity of 
proteins and polysaccharides: (1) physical methods electro¬ 
phoresis, ultracentrifuge, diffusion, solubility- ^which measure 
the effect of various forces on the molecule itself; (2) specific 
combining methods—activity (enzymatic, viral, hormonal) ancl 
antigenicity—which, to describe the properties of a given sub¬ 
stance, utilize the reactions between different molecules. In 
special cases, the chemical methods of analysis can be included 
here e.g., when the molecule under consideration has a known 
distinctive grouping, such as iodine in thyroglobulin, heme in cyto¬ 
These techniques for characterizing high molecular weight 
substances have become so refined that few purified proteins meet 
all the criteria of purity, for preparations apparently homogeneous 
by one test can be grossly Inhomogeneous by another. Ihis is 
not unexpected, since to a certain extent each method measures a 
different property of the molecule. The physical methods provide 

Uesesreli Council Merck fellow 111411-52. Pasteur tu.stitule, Paris. 





the only means for purification of proteins and polysaccharides, 
and the use of any particular technique selects those molecules 
whose homogeneity is assured only with respect to the property 
upon which the method depends. For example, a protein isolated 
by differential centrifugation would be expected to be homogeneous 
in the ultracentrifuge but might not be electrophoretically homo¬ 
geneous. Therefore, the least sensitive way of detecting different 
molecules within a given preparation would be to use the technique 
employed in the isolation. Thus the pure protein or polysaccharide 
must be defined in terms of all the techniques used to study it. We 
would call a preparation homogeneous when the diffevences be¬ 
tween molecules (ire less than can be detected by any of the known 


Although these methods are most otten consitlered only as tools 
for the detection of impurities, they also lead one to define a 
fundamental question of biolog}'' and, at the same time, they pro¬ 
vide the means to study it. The problem is to correlate physical 
and specific combining properties. Such a correlation involves the 
determination of which structural units are responsible for the 
specificity of a protein or polysaccharide, and of which physical 
properties ai’e relevant or irrelevant (and to what extent) to the 
given specificity. These problems are very difficult to investigate 
and are largely unsolved. For instance, tetanus and diphtheria 
antitoxins, which are easily separated by certain of their specific 
combining properties, have never been distinguished by physical 
methods. Furthermore, diphtheria antitoxin, a single biologic 
entity as regards its qualitative ability to combine with toxin, can 
be associated with protein molecules of widely different solubility, 
mobility and sedimentation properties (2). 

This section deals principally with an immunochemical method 
of characterizing proteins and polysaccharides, namely, the pre¬ 
cipitin reaction. Although many proteins do not have known 
measurable biologic activity, almost all are antigenic, and this pro¬ 
vides the basis for immunologic analysis. The most interesting 
and unique characteristic of this method is that one can prepare, 
at will, a specific test substance—antibody—for any given protein. 
Furthermore, this spec-ific reagent enables the establishment of 
an independent classification ot different molecules which are 
structurally more or less related. In addition, under certain condi¬ 
tions this analysis has extraordinary sensitivity, greatly exceeding 
t hat of the p hysical techniques for the detection of impurities, f 

“impurity,” when referring to imnmuoeheiui- 
saichSidS^ in the^Llw^ "“"t" ^^^tigenically distinguishable protein or poly- 


In short, the interest in the antigen-antibody reaction as a 
research tool stems from the desire not only to detect traces of 
impurity in purified proteins, but also ( 1 ) to assay for a given 
antigen in mixtures of unrelated substances, ( 2 ) to reveal and char¬ 
acterize structurally related (cross-reacting) proteins and ( 3 ) to 
study certain alterations in a molecule that might occur during 

The techniques of immunochemistry are not difficult and may 
easily be employed in any laboratory. Most of the methods and 
theory have been described by Rabat and Mayer (I), and the 
reader is referred to their excellent book for both the background 
and the bibliography of the subject. We do not intend to review 
the literature. Examples chosen are those most conveniently 
available to us, not necessarily the only ones. The section is de¬ 
signed essentially to provide details of techniques, but a discussion 
of the theory has been included because the interpretation of re¬ 
sults is the more difficult aspect of this type of work. Because 
of the essentially empiric nature of immunochemistry, there 
are wide differences of opinion, of interpretation and of experience. 
These will be indicated, and the reader asked to judge for himself. 

The review is divided into 3 chapters, each of which describes 
a particular aspect of immunochemical analysis. 

I. Immunization of the experimental animal. 

II. Fractionation of antibody containing 7 -globulins from vari¬ 
ous species. 

III. The techniques and analysis of the quantitative pre¬ 
cipitin reaction: A, the reaction in liquid media; B, the reaction in 

1 . 



:abat, E. A., and Mayer, M. M.: ExpenwentoZ Zwmunoc/iemisfrt/(Spring- 
field, III: Charles C Thomas, Publisher, 1948). r ,• • 

:ekwick P. A., and Record, B. R.: Some physical properties of diphtheria 
antitoxin, Brit. J. Exper. Path. 22: 29, i946. 



The preparation of high titer antisera is the key step in the use of 
immunologic methods. Despite this, the technique of immuniza¬ 
tion has been the least developed phase of immunology. Two ex¬ 
tensive analyses of the problem of antibody synthesis have ap¬ 
peared to which the reader is referred (2, 37). 

The investigations of Freund and collaborators (10-lG) suggest 
a new era of research which will lead to the production of repro¬ 
ducible preparations of antisera under absolutely controlled condi¬ 
tions. Since quantitative results on the immunization of animals 
are rare, we will discuss several practical aspects of the problem 
from the point of view of preparation of precipitating antibody to 
be used for studies on homogeneity of antigens. 


a) The animal .—Under ordinary conditions one will not find in 
the serum of animals antibody directed against their own proteins. 
There are certain exceptions, e.g., the organ-specific antigens—lens, 
testes. These antigens have not been isolated, and may not be 
proteins, but are antigenic in their own species (33). There are in¬ 
dications in the literature that antigens which cross-react with 
antibody against the proteins of a given species will, in certain 
cases, induce antibody formation in that species. For example, 
I hlenhuth (43) found that a hare could be immunized with rabbit 
senim protein and produce an antiserum very highly specific only 
for rabbit protein, whereas if a chicken were immunized, the anti¬ 
serum cross-reacted very extensively with rabbit and hare anti¬ 
gens. Another case is presented by Oudin who studied the cross¬ 
reaction between rabbit antisera to horse serum proteins, and the 
proteins of a wide variety of species, including man (35). He demon¬ 
strated that the serum albumin of man anti of horse cross-reacted 
with rabliit antibody to horse albumin; yet, as is knovm, human 
serum albumin is a good antigen in the horse (17, 42). Other cases 
are to be found in the work of Landsteiner (33). It is not known 
however, whether the antibody provoked by the closely related 
protein IS directed against the common structural configurations or 
ot. This important problem deserees further investigation, since 



it may enable the preparation of very specific antisera and may 
permit the study of specificity of antibody induction. However, in 
general, for investigations on homogeneity of antigens, one would 
avoid immunizing animals with their own or cross-reacting pro¬ 

Injection of a foreign substance into an animal does not guaran¬ 
tee an antibody response, for substances antigenic for one species 
may not be so for another. For example, isolated pneumococcus 
polysaccharide is antigenic for the mouse but not for the rabbit. 
This is an extreme case, but many substances have been described 
as good antigens in one species and poor antigens in another, and 
certain animals, e.g., guinea pigs, are considered poor producers of 
precipitin antibody while others, e.g., chickens, are considered good 
precipitin producers. In general, these statements are of limited 
significance for, by varying the route of injection or the dosage, by 
using adjuvants or by injecting the antigen coupled to another 
substance, the so-called poor producer of antibody may become 
a good one. For instance, if pneumococcus polysaccharide is in¬ 
jected as part of the whole bacteria cell or linked to a protein by 
artificial means, it becomes antigenic for the rabbit. 

Most often, one does not have clean-cut, all-or-none reactions to 
antigens. Certain animals seem to respond better to one group of 
antigens than do others. For example, the horse appears to be a 
good producer of antitoxins and the rabbit a poor one. However, it 
is difficult to generalize, as no controlled experimental data are 
available in which the 2 species are compared. It is not i^ssible to 
predict what substance will be antigenic m which animal. 1 he 
question at present can be solved only empirically. 

The animal’s history is important. Previous contact with an 
antigen will often result in very high titer antisera, rapidly pro- 
dueed on injection of n lecnll dose. This nnamnest.c response has 
been used to prepare hish titer diphtlieria antitoxin in both human 
beings and liorscs by giving injections to 

uals ((), 30). There does not seem to be an anamnestic respo . 
Dolvsaccharide antigens, a phenomenon discussed elsewheio ( ). 

" MaTum aLals^are u.ially better antibody than 

itrlv Sid in elucidating the problem of immunisation and would 
Carlinfanti (4). 



h) The antigen. —1) turity. —In geneml, a given antigen will 
elicit maximal antihnfly production when it is injected pure. If one 
immunizes with a mixture of .several antigens, the probability ot 
ha^'ing antibody against any given one decreases. For example, 
isolated cliicken ovalbumin is a fairly good antigen in the rabbit, 
however, whole egg white, which contains (iO per cent of this com¬ 
ponent, induces the production of much less antibody to ovalbumin. 
The major part of the antibody is directed against the very minor 
components of egg Avhite (46). In certain cases, even with purified 
antigens, more antibody is produced against the minor component 
than against the major one (7). Results of this kind could be inter¬ 
preted as meaning that the minor components, having greater anti¬ 
genic power than the major one, tax the limited capacity of the 
antibody-synthesizing mechanism to produce antibody against 
them, thereby inhibiting or limiting the synthesis of antibody to the 
major component. 

The response depentls in part on the dosage of immunization 
(see p. 276). However, this interference by minor components does 
not always occur, for there are (a) cases in which addition of un¬ 
related material enhances the antibody response, e.g., addition of 
staphylococcus toxin* (1), and (6) systems in which the antibody 
response to a given antigen is almost independent of its purity, 
e.g., diphtheria toxoid (26). 

2) STABILITY. —No clcar-cut data exist on the relation between 
the stability of a substance and its antigenicity. Yet it seems essen¬ 
tial that the structural and combining groupings of the molecule be 
stable enough to reach the antibody-forming centers before being 
destroyed. Altered proteins arc known to induce antibody of dif¬ 
ferent specificity from that induced by the native form (34). A 
similar phenomenon may be obtainefl with proteins or polysac¬ 
charides immediately destroyed or denatured when injected in the 
blood stream, e.g., pepsin (40). 

Antigens may be destroyed in the animals by various means, 
physical (maintaining at unfavorable pH and or temperature) and 
enzymatic (protcolytii* action). There are some indications that 
resistance of a protein to the animal’s proteases may play an im¬ 
portant role in fletermining its antigenicity. For example, the M 
protein of streptococci is readily destroyed by trypsin and is a very 
weak antigen, whereas the T protein is resi.stant to the protease and 
IS an excellent antigen. 

The question of stability may be more serious when the antigen 
iSMnjected as an emulsion, which is absorbed slowly over a period 
of weeks fr om a sterile abscess, during which time the protein mav 

* Sometimes irritants have a stimulating effect on antibody production. 


be modified. This problem is probably less serious with most poly¬ 
saccharide antigens which are generally very stable and in certain 
cases not known to be attacked by animal enzymes. This is re¬ 
flected in the general property of polysaccharide antigens to induce 
a level of antibody formation which remains high for long periods, 
in contrast to certain antiprotein antibodies, the levels of which, 
after the end of immunization, decrease more rapidly (23, 31). 

A protein might be stabilized in vitro and therefore possibly as 
an antigen in several ways, e.g., addition of certain metal ions 
(19, 30) or of the substrate in the case of enzymes (3); furthermore, 
by treatment with formaldehyde, the antigenic structures might 
be stabilized (26). For preparation of antisera against easily dena¬ 
tured or nonantigenic proteins, the last method might prove very 
important. Treatment of toxins with formaldehyde to produce tox¬ 
oids results in a loss of toxicity, but antigenicity remains unchanged. 
With respect to its antigenic properties, the resultant protein is 
much more resistant than the parent toxin to inactivation by heat 
or by extremes in pH. Treatment with formaldehyde might prove of 
great value in rendering weakly or nonantigenic proteins like the 
M substance antigenic. For example, when papain, which is not 
antigenic for the guinea pig, was treated with formaldehyde, it lost 
enzymatic activity but induced formation of antibody which neu¬ 
tralized papain activity (38, 48). j- u.i • 

Holt (26) has described a method of formahnizmg diphtheria 

toxin that seems to be generally applicable. 

There are certain groups of substances considered non- or weakly 
antigenic. Low molecular weight proteins (protamine, gelatin, 
lysozyme) (39, 44) and glycoproteins (ovomucoid) (45) chondro- 
mucoid (9, 18) have been said to be weak antigens. It is difficult t 
explain these findings on any single basis. The ^ 

glycoproteins might be a phenomenon specific for the '^hbit, and 
fhe behavior of low molecular weight proteins might be accounted 
for in terms of their rapid excretion* or destruction. Use of adji - 

vants or treatmentwith formaldehyde ® ^fdv r^- 

highly antigenic. In the case of lysozyme, an enhanced antibody re 

spouse in rabbits has been obtained with use of 
ho?L are most useful for preparation of 



chicken is a good producer of precipitating antibody and is ex¬ 
tremely hardy. However, for quantitative work, its serum should be 
fractionated because of an apparent nonspecific co-precipitation o 
certain a-globulins (8). The horse, the only kno^^m producer oi 
flocculating-type antiseraj (see p. 323), is extremely useful for 
antibody production, but good facilities for injection and upkeep 
(as well as large amounts of antigen) must be available, thus 
limiting its use. 

In the case of toxic antigens, apart from use of formaldehyde to 
render them nontoxic, it is sometimes possible to immunize a rela¬ 
tively resistant animal, i.e., urease in the hen (41). 

4) PREPARATION OF ANTIGEN. —For most proteins whose anti¬ 
genic properties are unknown, the adjuvant technique of Freund 
et al. (14) is most convenient. The antibody response is often greatly 
enhanced. This procedure has the advantage of using less antigen 
and fewer injections. The Freund technique follows. 

For the emulsion, materials include; 8.5 vol of Bayol F, a light 
paraffin oil (Stanco Distributors, New York) with 10 mg heat- 
killed Mycobacterium§ per 100 ml; 1.5 vol Arlacel A, a mannide 
mono-oleate (Atlas Powder Co., Wilmington, Del.); 10 vol of the 
antigen in buffer. 

The Bayol F and Arlacel A can be sterilized by autoclaving; 
the antigen can be preserved with Merthiolate 10~^ by weight or 
sterilized by filtration. The heat-killed (70 C, 30 min), dried, acid- 
fast organisms, e.g., Mycobacteria tuberculosis or butyricum, are in¬ 
corporated in the Bayol by grinding bacteria and oil together in a 

The mixture of Bayol F and Arlacel A is made, then the aqueous 
phase added slowly during mixing. A Waring Blendor, with speed 
rheostat controlled, may be used. If possible, the Blendor should be 
kept cold to prevent overheating of the emulsion. With the small 
metal container supplied for the Blendor, as little as 5 ml of solu¬ 
tion of antigen can be emulsified. If smaller quantities are to be 
prepared, or if a Blendor is not available, the mixture can be worked 
up and down in a syringe without needle, ground in a mortar or 
stirred with an electric-motor-driven bent glass rod. To test the 
emulsion, a drop at the end of an applicator stick is allowed to fall 
about 5 cm onto the surface of water. If the drop remains perfectly 

t brief, the reaction of flocculating antibody is characterized bv the solubilitv 
of the antigen-antibody complex in excess antibody and a broS equivalence 


formed and does not spread out over the surface, the emulsion is 
ready for injection. 

Alycobacteria often augment the antibody response. Although 
one must be careful, vheii dealing with antigens from bacterial 
sources, to verify that no antigens of the Alycobacteria cross-react 
with the new antigen, it is advisable to immunize several rabbits 
with and without the Mycobacterium to observe its effect. 

The QUANTITY OF ANTIGEN for injection by the Freund technique 
depends on whether the resultant antibody is to be used for assay 
purposes or to detect impurities. The production of antibodies 
against minor components is fa^'ored by giving massive doses over 
a short period of immunization, e.g., 50-100 mg antigen/rabbit 
and bleeding after 3 weeks. Smaller doses, 0.5-5 mg antigen/rabbit, 
and 6-8 weeks of immunization favor the production of antibody 
against the major component. We have had excellent antibody re¬ 
sponse in rabbits, guinea pigs and monkeys with single injections 
of as little as 0.100-0.500 mg diphtheria toxoid (6), ovalbumin or 
jS-galactosidase (5). Grabar et al. (21) noted an excellent antibody 
response in rabbits with as little as 0.1 /xg typhoid O antigen. In 
man, Heidelberger ei al. have shown that as little as 30-40 gg 
pneumococcus polysaccharide is actively antigenic (25), and as a 
recall dose in man, 0.4 ng diphtheria toxoid can provoke an enor¬ 
mous response (6) of the order of 10® molecules of antibody for each 
molecule of antigen injected. For the horse, larger doses might be 
necessary, as much as 2-5 g of antigen/500-700 kg animal. 

Several other carriers of antigen have been described, of which 

protamine, aluminum hydroxide and aluminum phosphate ap¬ 
pear to be the most useful. Holt (26) has described the use of the 
latter 2 sulistances for immunization against diphtheria toxoid 
(see his book for description of the properties and preparations of 
these carriers). About 20-30 mg of antigen per 3 kg rabbit, divided 
into about 20 injections, is the usual dosage with this technique. 

In the case of certain polysaccharide antigens, imrnunization 
of the rabbit is carried out by injecting whole organisms. The 
pneumococcus polysaccharides can seiA'e as an example. le 
organisms may be grown on neopeptone meat infusion broth, con¬ 
centrated by centrifugation and resuspended in physiologic saline 
in a concentration of about 0.05 mg bacterial nitrogeu/ml Ihey 
are then heat-killed at 65 C for 30 min in a water bath or trea ed 
with 0 2-0.5 per cent Formalin. Since the polysaccharide of t e 
gram-negative form loses antigenicity, they must ^ 

Gram stain before injection. Freund and Hehan report that rabbit 
serum against the polysaccharide of type III pneumococcus can be 
prepared with living pneumococci more readily than by using i ( 


ferial cells killer! by the usual methods (Formalin, flash heat, etc.) 

^^5) ROUTE AND SCHEDULE OE INJECTIONS.— With most animals, 
and with the antigen prepared in adjuvants, several techniques are 
possible. In general, a single injection of the antigen emulsion is all 
that is necessary before the test bleeding. For rabbits, the most 
satisfactory site of injection is the space behind the scapula, be¬ 
cause it cannot be irritated by the animal. For chickens and guinea 
pigs, intraperitoneal injections are best, although the subcutaneous 
route is also effective. Three to 4 weeks after the beginning of im¬ 
munization, a test bleeding is made. If the titer is not high enough 
(see p. 279), another injection with or without adjuvants often 
produces an enhancing effect, and further test bleeding is made 3 
weeks later. If antibody is formed, its titer in the serum is usually 
sustained at a high level for many months and at a lower level for 
several years (I6a). If no antibody is found after this, the other 
suggestions already discussed can be tried. 

For antigens adsorbed on gels, in solution or in bacterial suspen¬ 
sions, a series of intravenous injections followed by subcutaneous or 
intraperitoneal injections is most often used. In a typical schedule, 
the animal is given I intravenous injection a day for 4 days and al¬ 
lowed to rest for 3 clays. This may be continued for 4 weeks. To 
avoid danger of anaphylactic shock, it is sometimes necessary to 
change the route of injection, and after the 1st week the antigen 
preparation is given subcutaneously or intraperitoneally. 

Another variation of the schedule is to give injections daily 5 
times a week, the 1st subcutaneously (to desensitize and protect 
against anaphylaxis) and the following 4 intravenously. It is well 
to increase gradually the amount injected, starting with 1 mg total 
antigen/injection and increasing the dose to 5 mg. 

The experience of workers with the methods of immunization 
differs, and each laboratory holds strongly to its own techniques. 
I have described the Freund technique in detail because it has 
proved successful with a variety of antigens and because very small 
amounts of antigen are necessary for maximal response—an im¬ 
portant factor in work with precious materials. 

However, Kabat and Mayer prefer to use repeated intravenous 
injections of alum-precipitated antigens for rabbits, because they 
have found much higher antibody response with ovalbumin, 
human serum albumin, y-globulin and chymotrypsin as antigens. 
Their technique is described in their book (34).^If sufficient of an 
untried antigen is available, it is recommended that 2 groups of 
animals be used, 1 with alum-precipitated antigen and the other with 
adjuvants. For guinea pigs, the latter technique is preferable. 


JMany observations have been made that the route of injection is 
important in the production of antibody (reviewed by Grabar 
(20)). From available evidence it seems that for the horse, the anti¬ 
body response to protein antigen is minimal when intravenous in¬ 
jections are used. This is not true for polysaccharides. For other 
species, although the route of injection is probably important, the 
data do not justify practical conclusions regarding the production 
of antisera. The reader is referred to a series of fine investigations 
on the antibody response of rabbits and guinea pigs to diphtheria 
toxoid, as a function of dosage, route of injection, carrier of antigen 
and secondary response (15, 10, 26-29). 

Careful study of all the available information on methods of im¬ 
munization for commercial production of antibodies in horses 

shows that many different procedures are used successfully. The 
antigen in solution adsorbed on alumina gels or in adjuvants is in¬ 
jected by combination of all known routes—intravenous, intra- 
peritoneal, intramuscular and subcutaneous—with total quantities 
of antigen administered ranging from 1 to 5 g in horses weighing 
500-700 kg." 

There is no general procedure, as far as I know, which can be 
recommended with certainty for the production in the horse of 
floccvlalinq antibody to untried protein antigens. The data on im¬ 
munization to diphtheria toxoid are not generally applicable be¬ 
cause most workers choose Schick-negative horses and take ad¬ 
vantage of the booster response. For serum albumin, Gitlin et al 
(17) described a procedure beginning with alum-precipitated anti¬ 
gen and finishing with adjuvants. In its essentials, their technique 
Ls been repeated by Grabar at the Pasteur Institute, h the 
same antigen in the horse, and very high titer flocculating antibody 
was obtained. Possibly their immunization program is generally 
anplicable. I have had some experience in immunizing large animals, 
donkev goat and sheep, with chicken ovalbumin, human 7 - 
globulin atoxic diphtheria proteins and ^-lactoglobuhn. Responses 
Ire exceVnt when the procedure outlined below was followed 

These animals produced precipitin type Tot- 

be certain that the procedure will carry over to the horse. Hon 

ever there is sufficient chance that it will be useful and 

.ibt, c”. .»'l' •■"t “ 

institutes producing antitoxins. 


mg total antigen and increasing it 10 mg a week for 8 weeks.* 
Make test bleedings and examine serum for nonprecipitatmg as 
well as precipitating antibody. Presence of nonprecipitating anti¬ 
body in the early stages of immunization is a good sign that a 
flocculating antibody will appear (9, 33). If no response is found, 
rest the animal for 1 or 2 months and repeat immunization. It anti¬ 
body is formed, continue injection every 2 weeks with 150 mg anti¬ 
gen, making frequent test bleedings to see if the titer increases. For 
polysaccharide antigens, the intravenous route is preferable, with a 
schedule similar to that used for the rabbit, except that a dose of 
20-50 times as much antigen is recommended. 

ings are extremely important and should be carried out carefully 
(for techniques of bleeding animals, see Kabat and Mayer (32)). 
Before injections are given, a control test bleeding must be made, 
and it should be shown that the given serum of the animal species 
does not react nonspecifically with the antigen to be studied. 

After immunization, the antisera are screened by a ring test. 
The antigen in buffer solution (about 0.3 ml) is carefully layered 
over the antiserum (0.3-0.4 ml) in a narrow tube 3-4 mm in di¬ 
ameter. Appearance of turbidity or precipitation at the interface 
constitutes a positive result. Sera giving a negative reaction can be 
rejected. The precipitating sera can then be tested semiquantita- 
tively. To 1 ml of each serum add about 100 jug of antigen. The 
precipitate is centrifuged off after 1 hr at 37 C. Another 50 ng of 
antigen is added to the supernatant and gently mixed without 
disturbing the precipitate. If further precipitation occurs after 1 hr 
at 37 C, the process is repeated until further addition of antigen 
causes no precipitation. The mixtures are allowed to stand over¬ 
night in the icebox (0-4 C), the precipitates are washed and their 
nitrogen contents determined (pp. 301 ff.). In ordinary circum¬ 
stances, antisera containing less than 200 /xg antibody N/ml should 
not be used.f 

Animals showing good antibody levels should be exsanguinated 
after a fast of 18 hr to obtain sera free from dissolved lipids which 

f The use of subcutaneous injections is indicated also because some preliminary 


add screening precipitin-type antisera is to 

antigen N to 1 ml portions of serum and, after incubation at 5 C 
overnight, separate the supernatants by centrifugation. These supernatants are 
determine whether there is antigen or antibody excess (p. 307). If there is 
antibody excess after precipitation with the 25 iig N the antiserum is nrobnbW 


separate on standing. It seems to this worker that prolonged im¬ 
munization (6-12 months) and resultant frequent bleeding of the 
animal should be avoided hecause (1) the chances of having anti¬ 
body production against minor impurities increase, (2) prolonged 
immunization may lead to a broadening of the equivalence zone 
and increasing ratio of antibody/antigen (24), and (3) in general, 
long-continued immunization favors the production of sera with 
less specificity. For the study of homogeneity, these effects would 
complicate the analyses and make any possibility of interpreting 
the ratios (antibody/antigen) in terms of molecular weight mean¬ 
ingless. However, for certain purposes, such as use of antiserum 
for assay and with very pure antigens, the continued immuniza¬ 
tion of rabbits offers no difficulty and in fact allows a saving on 

For most work, I believe that sterile technique is not necessary 
and is, in fact, cumbersome. The blood is collected in 50 ml Luster- 
oid plastic tubes and allowed to clot at room temperature. The clot 
is then separated from the wall (ringed) with an applicator stick 
and the tube placed at 37 C for 1 hr to let the clot retract. The 
serum is separated in an angle centrifuge for 1 hr at 5,000 rprn, col¬ 
lected and pooled, t The serum is allowed to stand overnight in the 
icebox and recentrifuged. Merthiolate lO"'* parts by weight and 
0.2 per cent phenol are satisfactory preservatives provided they do 
not affect the antigen. 

Comment by E. A. Rabat 

It is the invariable practice in our laboratory to collect an(l process 
bloods sterilely and to add preservative. IMuch of the analytical work is 
also done with sterile glassware except for analytical pipets which cannot 
be heated without risk of altering their calibration. However, cotton plugs 
are put into the mouth end before use to avoid gross contamination. 1 ese 
precautions are essential in working with small quantities of antibody when 
tubes are allowed to remain in the refrigerator for a week before analyses 
are carried out and especially when repeated set-ups are made on the supei- 
natant It is also essential when sera ai e to be used foi- passive transfei es 
"an beings; for such studies the use of M per cent phenol .s even 
lireferable since it has considerable anesthetic effect on the s m. 

Plasma is not very satisfactory as a source of antibody, because 
fibHnTrtTpffLes tend to form. The serum can be s^atisfactonly 
kLt in solution or, in the case of T-glohulins, can be stored at 
—10 C as the lyophilized powder over P 2 OS in vacuum. 
sera° Heidelberger and Pi Lapi (22) present evidence that the best 

1 Use'of planLic tubes iiistejid of kIjws is oiyuiusfbe careful, witli 

Sat;;'s'«li: S: S does not oai«e beiuo.vsis 

(iue to destruction of red [flood cells. 



way to conserve the antibody is to store it in the frozen state at 
solid CO 2 temperature or simply in solution at 5 C. 


1. Burky, E. L.: The production in the rabbit of hypersensitive reactions to 

lens, rabbit muscle and low ragweed extracts by the action of staphylo¬ 
coccus toxin, J. Allergy 50: 466-475, 1934. 

2. Burnett, F. M., and Fenner, F.: The Production of Antibodies (2d ed.; 

New York: Macmillan Company, 1949). 

3. Burton, K.: Stabilisation of D-aminoacid oxidase by flavin adrenine 

dinucleotide substrates and competitive inhibitors, Biochem. J. 48: 

458-470, 1951. 

4. Carlinfanti, E.: Problemes quantitatifs de I’immunitd, Ann. Inst. Pasteur 

79:351-372, 1950. 

5. Cohn, M., and Monod, J.: Purification et propri4t6s de la b6ta-galacto- 

sidase d'Escherichia coli, Biochem. biophys. acta 7: 153-180, 1951. 

6. Cohn, M., and Pappenheimer, A. M., Jr.: A quantitative study of the 

diphtheria toxin-antitoxin reaction in the sera of various species includ¬ 
ing man, J. Immunol. 63: 291-312, 1949. 

7. Cohn, M.; Wetter, L. R., and Deutsch, II. F.: Immunological studies on 

egg white proteins: I. Precipitation of chicken-ovalbumin and conal- 
bumin b}’- rabbit and horse-antisera, J. Immunol. 61: 283-294, 1949, 

8. Deutsch, II. F.; Nichol, J. C., and Cohn, M.: Biophysical studies of blood 

plasma proteins: XI. Immunochemical and electrophoretic studies of 
immune chicken serum, J. Immunol. 63: 195-206, 1949. 

9. Elliott, C. H.: The antigenic properties of glycoproteins, J. Infect. Dis. 

15: 501-511, 1914. 

10. Freund, J.: Some aspects of active immunization, Ann. Rev. Microbiol 

1: 291-308, 1947. 

11. Freund, J., and Bonanto, M. V.: The effect of paraffin oil, lanolin-like sub¬ 

stances and killed tubercle bacilli on immunisation with diphtheria 
toxoid and Bact. ti/phosum, J. Immunol. 48: 325-334, 1944. 

12. Freund, J., and Walter, A. W .: Saprophytic acid-fast bacilli and paraffin 

oil as adjuvants in immunization, Proc. Soc. Exper. Biol & Med 56- 
47-50, 1944. ■ ‘ ■ 

13. Freund, J., et ai: ,\ntibody formation and sensitization with the aid of 

adjuvants, J. Immunol. 60: 383-398, 1948. 

14. Freund, J., and Behan, M. A.: Type-specific antibody production with 

living pneumococci in the rabbit. Science 90: 185-186, 1939 

15. Freund J., and Bonanto, M. V.: Antitoxin formation after intravenous 

or subcutaneous injection of plain or alum diphtheric toxoid J Im¬ 
munol. 40: 437-447, 1941. 



17 . 


reund, J., and Bonanto, M. V.: Effect of the amount of antigen on anti¬ 
toxin formation during the inimary and secondary immunisations J 
Immunol. 45: 71-82, 1942. ‘ ’ 

Freund, J., and Bonanto, M. V.: The duration of antibody-formation 

mull 5l:'2Ml4ri94r‘'"'^ J- 

Gitlin, D., Davidson, C. S., and W'etterlow, L. II.: The quantitative e^ti 
mation of serum albumin in human body fluids by direct t itration with 
specific antiserum, J. Immunol. 63: 415 1949 

37:'285,'Kl'lcopn.Voins, J. Infe.1. DU. 



20 . 
21 . 

22 . 





















Gorini, L.; Role du calcium dans le systeme trypsine-s6rum albumine, 
Biochem. biophys. acta, 7: 3-8, 1951. 

Grabar, P.: Immunocliemistry, Ann. Rev. Biochem. 19: 453, 1950. 
Grabar, P., and Oudin, J.; Etude sur le polyside “O” d’un serum de lapin 
sp6cifique de ce polyoside, Ann. Inst. Pasteur 73: 627, 1947. 
Heidelberger, M., anti DiLapi, M. M.: Measurement and preservation of 
antibodies in human sera, J. Immunol. 61: 153, 1939. 

Heidelberger, M., el al: Persistence of antibodies in human subjects in¬ 
jected with pneumococcal polysaccharides, J. Immunol. 65: 535, 1950. 
Heidelberger, M., and Kendall, F. E.: The reaction between crystallinjj 
egg albumin and its homologous antibody, J. Exper. Med. 62: 697, 

1935. . . r „ . . . r 

Heidelberger, M., et al.: Antibody formation in men following injection ot 
pneumococci or their type specific polysaccharides, J. E.xper. Med. 83: 

303, 1946. , , . T . AVIV 

Holt, L. B.: Development in Diphtheria Prophylaxis, London, William, 

1950. . , , • n.. 

Holt L. B.: Quantitative studies in diphtheria prophylaxis: ihe primary 

response, Brit. J. Exper. Path. 30: 289-297, 1949. . ^p. „ j 

Holt L. B.: Quantitative studies in diphtheria prophylaxis: The second 
response, Brit. J. Exper. Path. 31: 233-241, 1950. _ 

Holt L B.: Quantitative studies in diphtheria prophylaxis. Some o - 
servations on the correlation in identified guinea pigs between then- 
responses to more than one inoculation of diphtheria prophylactic, 32. 

HuRon^^E!: On the stability of alpha-amylase by lack of calcium, Arkiv 
Kemi Mineral. Geol. 2: 135-145 and 173-187, 1950. 

Jensen C : Antitoxin curves in children after active immunisation \uth 
"Sheria anatoxin, with special reference to ‘he duratron of antrtoxrc 
irnTTiiinitv Acta path, et microbiol. scandinav. 10. 137 151, 

Rabat, E. A., and Mayer, M. M.t ExpenmenM Immunochemtstry, 
ff?nrinafield Ill.: Charles C Thomas, Publisher, 1948). 

Landrtetae?, K.: SpecificUp of SerolopM Reacl.ons (Cambridge, Mass.. 

Harvard University 'J' , Preparation and immuno- 

tion au s«rum de cheval et an .ut de 
Parn^ei — in the horse. I. 

Pa'p^SheS: SbilS-: Nt 

S reXr;oSne, ^ Immunol. 9: ,34- 

Ramon, G.: Ferment, anaferment, antitermeiu, ixev 

Smotnb!'and Charney, J.: The antigenicity of crystalline lysosyme, J. 

Bact. 54: 101-109, 1947. Immunological studies on pepsin 

Seastone, C. V.. and Herriott NU • 

and pepsinogen, .1. Gen. P . - ■ „ o: 201-225, 193i. 

Sumner, .1. B.: Antiureiwe, Ui-Keta- U ^ Antiproteins in horse 

^^frlTil": InSLSuti albumin and their reaction with 


antigen; IV. Antibodies to rabbit serum globulin and their interaction 
with antigen, J. Exper. Med. 86: 83 and 95, 1947. 

43. Uhlenhuth, Von S.: Ein Verfahren zur biologischen Unterscheidung von 

Blut vervanter Tiere, Deutsche med. V chnschr., 1905. 

44. Wetter, L. II.: “Studies on Egg White and Blood Serum Proteins,” Doc¬ 

toral thesis. University of Wisconsin, 1950. 

45. Wetter, L. R., and Deutsch, II. F.: Immunological studies on egg white 

])roteins: III. Quantitative immunochemical studies of ovomucoid. 
Arch. Biochem. 28: 399, 1950. 

46. Wetter, L. R.; Deutsch, H. F., and Cohn, M.: unpublished experiments. 

47. Wilson, G. S., and Miles, A. A.: Topley and Wilson’s Principles of Bac¬ 

teriology and Immunity (3d ed.; Baltimore: Williams & Wilkins Com¬ 
pany, 1946), pp. 1091-1092. 

48. Zanco, M., and Pratesi, G.: Sur les r<5actions immunologiques 5, la pa- 

paine. Boll. Soc. ital. biol. sper. 23: 419, 1947. 


HAROLD F. DEUTSCH, University of Wisconsin 

It is often more desirable to study the chemistry of immune 
reactions with the “so-called” 7-globulin fractions than to use 
the whole serum. This allows the investigator to avoid or regulate 
the effects of non-7-globulin factors such as complement and 
lipids which are known to react with antigen-antibody systems. 
Furthermore, after separation of the 7-globulins from weak immune 
serum, an antibody system of much greater potency can be 
reconstituted. The separation of the antibody from such entities 
as serum proteases also makes it possible to store solutions of these 
proteins with less danger of deterioration than is the case with 
whole serum. 

Although reasonably good antigens may be obtained, the prepa¬ 
ration of pure antibody is difficult and somewhat controversial. 
A part of this difficulty is concerned with variations in mass and 
charge of the antibody molecules in serum. Thus antibody fractions 
of varying solubility, mass, electrophoretic and biologic properties 
may be obtained from the 7-globulin fraction of a given serurn. 
However, preparations showing reasonably uniform electrophoretic 
properties may be readily obtained from immune serum, and while 
they in general usually contain less than 10 per cent specifically 
precipitable protein they do represent material devoid of the other 

recognized electrophoretic entities of the serum. , , 

The aqueous-ethanol fractionation techniques of Gohii et al. (dj 
have proved very suitable for the separation of such materia 
from immune senim. They have many advantages over neutral 
salt fractionation methods and consequently we will deal o" -V 

separation of antibody-active protein : 

We are thus also neglecting a consideration of the subject ot ant 
bodv oreoaration by the dissociation of specific precipdates. 

DeveC by Cohn al. (4) of more specific fractionation 
techniques which use metals as specific complexing agents wil 
undoubtedly provide better techniques for antibody sepaiationr 
r tliffuTure However, their applications to various animal sera 
;"l{ entail extensive investigation before practical procedures can 

be developed. 




Tl.e variables in the ethanol type of fractionation are the alcohol 
concentration, ionic strength, pH, temperature and 
centration. Their effects will be treated briefly and individually 
as regards their application to antibody fractionation. 

ALCOHOL.— Ethanol has been most widely used, although 
other alcohols or water-miscible organic solvents of low dielectric 
constant may be employed. However, it must not be construed 
that a given concentration of ethyl alcohol is synonymous m its 
precipitating effect witli another alcohol. Ethanol concentrations 
are usually expressed in volumes per cent or in mole fractions. 
The volume contraction resulting upon the addition of ethanol to 
an aqueous system may be neglected in the usual fractionation. 
Thus if 10 vol of 95 per cent ethanol is added to 15 vol of a 1 per 
cent protein solution the ethanol concentration is calculated as 
38 per cent or as 0.152 mole fraction. 

Ethanol under certain conditions is a good denaturing agent. 
Although various proteins show markedly different labilities in this 
respect, the antibody-active proteins are readily inactivated in 
the presence of alcohol if adequate temperature conditions are 
not maintained. The ethanol solutions are always precooled 
(-5 to -20 C) before addition to the protein system. The 
latter is cooled to near its freezing point, and in practice it is often 
desirable to have a small amount of ice in the system before the 
addition of ethanol. The temperature of the protein solution is 
lowered as the alcohol is added, usually being maintained within 1° 
of its freezing point until —5 to — 8 C has been reached. At this 
temperature the antibody-active proteins of serum in alcohol up 
to ethanol concentrations of 40 per cent appear to show little de- 
naturation as judged by the criteria of biologic activity and solu¬ 
bility in dilute salt solutions. 

When adding alcohol to an aqueous protein system it is best 
to start with less concentrated reagent. In practice, a 50 per cent 
solution of ethanol is added until a concentration of 15-20 per cent 
has been obtained, and further additions may be made with 95 
per cent ethanol. Use of the more concentrated alcohol obviates 

the obtaining of unnecessarily large volumes, but if this is not a 
consideration, more dilute solutions of alcohol may be used with 
somewhat greater safety. Obviously the rate of addition, the effi¬ 
ciency of stirring and the effectiveness of the agency for lowering 
the temperature during the ethanol addition are important factors 
m the fractionation. The temperature must fall as alcohol is added 
to an aqueous system, and the addition should not be carried out in 
such a fashion as to cause massive local precipitation with the in- 


elusion of much protein that should be soluble at the given pre¬ 
cipitation step. 

A point of some interest is the possible deleterious effects of 
ethanol. The 7 -globulin fractions of various animals have been 
found to contain varying amounts of material which appears to 
be approximately twice the molecular weight of the major portion 
(15). However, freshly fractionated (13) or electrophoretically 
separated (14) material appears to contain little or none of this 
heavier component. Cann et al. (2) have found that 7 -globulins 
prepared by electrophoresis convection behave similarly. Also, 
Heidelberger et al. (9, 12) have shown that horse antiprotein serum, 
when delipidated or fractionated by ethanol, may give material 
of apparent higher antibody titer as measured by quantitative 
precipitin techniques. This may be due to aggregation of anti- 
body-active and immunologically inactive protein in the 7 -globulin 
fraction to give material which still reacts immunologically. 
Such considerations need further investigation and should serve 
to remind the investigator that fractionation conditions may lead 
to the production of material and subsequent immunologic reac¬ 
tions which have no counterpart in nature. 

HYDROGEN ION CONCENTRATION. —The pH at a given alcohol 
concentration and temperature is not known with certainty. It 
is usually expressed as the value measured ( 1 ) at room temperature 
after dilution with weak NaCl solution, (2) at room temperature 
without dilution or ( 3 ) in the presence of the ethanol at a low 
temperature. The last condition is used when small amounts of 
valuable material must be fractionated and the amount of sample 
required for a pH determination cannot be discarded. The important 
consideration is to express under what conditions a given pH was 
obtained so that they can be met by another worker or m a future 

The solubility of a given protein under stated conditions o 
ethanol, salt concentration and temperature is markedly influenced 
bv the pH. It is usually at a minimum at the isoelectric point ot 
the protein. Furthermore, at a given pH some proteins ^^y show 
a net negative charge while others show the opposite. This may 
obviLe some of the usual factors of solubility because of comple 

formation, particularly if the salt concentration IS low. 

temperature.— The importance of temperature control 
terms of protein denaturation in ethanol systems ha® been men¬ 
tioned A 2d important consideration is that of solubility, sua y 
pronounid Sases in solubility are mediated by tempera^ 
ZereZs and vice versa. It is thus important to record the tem 
perature kt which a given precipitation was carried out, or van 


tions in temperature from 3 to 5 C may often result in marked 
changes in yield or an increase in the amount of an impurity, 

depending on the particular fractionation step. .11 • .i 

IONIC STRENGTH.— This represents an important variable in the 
ethanol fractionation; large changes in solubility may be effected 
as a result of relatively small changes in salt concentration. 
jMoreover, specific anion and cation solubility effects aie being 
recognized but have not been extensively explored. Our considera¬ 
tions will be limited simply to a consideration of ionic strength. 
It is assumed that the aqueous dissociation constants of the acids 
used more or less obtain in aqueous-ethanol solutions. The 
most commonly used salts in this type of work are those of acetate, 
carbonate, chloride and phosphate. In the usual pH range of 
fractionation (4-8), only the effect of the salt need be considered 
in the ionic strength (ju) calculation in which 

M = I S 

where m is the molality and v the valence of the ions in question. 
Since relatively dilute solutions are used the molarity may be 
readily substituted for the molality. It is apparent that the ionic 
strength contribution of any monobasic salt is the same as the 
molarity, but for a salt such as Na 2 HP 04 the ionic strength is 3 
times the molarity. Calculation of the ionic strength necessitates 
the use of the foregoing equation, the pH and the Henderson- 

r\o op 

Hasselbach relation of pH = pK -\- log ——. Thus if a known 


volume of protein solution is adjusted to pH 4.5 by the addition of 
dilute acetic acid or acetate buffer, the amount of the added buffer 
or acid which exists as sodium acetate may be readily calculated 
it the pH of acetic acid is known. The ionic contribution of the 
protein and acetic acid may be neglected. The ionic strength 
contribution to the system by the added salt is most readily deter¬ 
mined in practice by construction of suitable graphs which show 
the relation between ionic strength and molarity at a given pH. 
In Figure I such a nomogram for acetic acid and phosphoric acid 
for the pH range 4-8 is given. The pH of acetic acid was taken as 
4.73. Only pKg, of the phosphate system, need be considered over 
this pH range and was taken as 6.7. 

In initiating an antibody fractionation with serum it is con¬ 
sidered that this system has an ionic strength of 0.15. Since in 
our practical preparation methods the 1st precipitation is carried 
out under conditions where the serum undergoes 8-fold dilution 
the ionic strength now lies in the region of 0.02. The initial crude 


antibody precipitate is usually about 25 per cent protein and is 
suspended in water to give a 1-2 per cent protein solution for the 
final purification step. This results in an ionic strength of 0.0015- 
0.0008 before adjustment of the pH. After adjustment, the ionic 
strength may be calculated and then brought to the desired level 
before the alcohol addition by use of a buffer solution of the pi’oper 
pH 01 of a neutral salt such as NaCl. Since the ionic strength is 
expiessed in terms of the final precipitating condition, the diluting 
effects of the alcohol must be considered in this adjustment. The 


\—Ionic strength and pH relationship of acetate and phosphate buffers. 

requisite amount of salt must have been added to the piotein 
solution before the addition ot alcohol in most fractionation steps. 

PROTEIN CONCENTRATION. —Ill theory it is desirable to frac¬ 
tionate extremely dilute protein solutions, but solubility considera¬ 
tions effect a more practical approach. If undiluted serum is used 
the liquid occluded in the precipitate will carry more contaminat¬ 
ing protein than if a diluted system was utilized. A practical example 

of this follows. 

Example: If a 1:4 dilution of serum is brought to a concentration of 25 
per cent ethanol by the addition of 50 per cent ethanol, a 1:8 dilution o 
the serum has been effected. If for each 100 ml of a serum of 7 per cent pm- 

tein concentration, 1.5 g of protein is precipitato “ ..''""T n 

cent protein paste, the solvent in the precipitate will contain appioxi- 



rr of soluble protein. Had the initial precipitate been removed 
by atdiUon of ireXl U'me of 50 per cent ethanol to the undJutel 
smim an equal amount of precipita te (Log) would contain approximately 
0 12 g’of occluded protein. Actually the lower volume resulting in the case 
of undiluted serum would probably result in more protein being oc¬ 
cluded than indicated Iiecause of moi-e protein-protein interaction, pre¬ 
cipitation of more protein that would be soluble m less concentrated solu¬ 
tion etc. The actual protein concentration employed in a fractionation is 
thus an empiric consideration in light of our present knowledge and ( e- 
pends on such factors as type of serum, amount being fractionated de¬ 
sired purity and nature of the material being isolated. In antibody frac¬ 
tionation work we have usually utilized precipitating conditions which 
employ approximately 1 per cent protein. 


General considerations .—Any quantity of antiserum may be 
fractionated. It is imperative that a refrigerated centrifuge or a 
refrigerated room be available. Low temperature must be main¬ 
tained during centrifugation. The precipitates obtained may be 
treated in various ways. If they are to be subjected to further 
fractionation they may be stored as the alcoholic pastes for some 
time at temperatures near their freezing points. It is, however, 
better to complete the fractionation as rapidly as possible and 
to avoid any such storage period in the form of an ethanol paste. 
During dissolving or suspending of any ethanol precipitate for 
lyophilization or further fractionation it is desirable to add to 
the pastes a small quantity of crushed ice before addition of the 
cold solvent. This tends to keep the temperature at a minimum 
during the suspending process. Small amounts of precipitate may be 
dispersed with a glass rod or a test tube type of homogenizer. Larger 
amounts may be suspended with a Waring Blendor or a similar 
device. Pastes representing the desired antibody fraction may be 
preserved for use in various ways. They may be suspended in 
water or dilute salt and then lyophilized. The use of a small amount 
of salt in this operation appears to give more stable protein prepa¬ 
rations. The protein pastes may also be dissolved directly in an 
ice-salt or buffer solution and then dialyzed in the cold to remove 
the alcohol in the precipitate. This latter procedure tends to give 
more stable protein solutions than those subjected to lyophili¬ 
zation procedures. 

The detailed procedures to be described for preparation of the 
antibody-active fractions ol the serum ot various animals are those 
with which we have had experience. Each system will be discussed 
briefly and individually and certain problems specific for a given 
serum will be treated. 


The use of serum as starting material for preparation of 7 - 
globulin is recommended because in the methods discussed here 
small amounts of fibrinogen are sometimes carried over to the 
7 -globulin fraction when plasma is employed. In solution this 
protein tends to flocculate slowly or clot out of solution, making 
it difficult to maintain a standard antibody solution suitable for 
quantitative immunochemical investigations. If plasma is collected 
it is recommended that the fibrinogen be removed before fraction¬ 

INITIAL PRECIPITATION. —The antibody-active proteins of all 
of the animal serums studied to date represent those electro¬ 
phoretic globulin entities of the serum of highest isoelectric point. 
For this reason an identical initial precipitation step may be 
carried out for all the different serums. This consists in diluting 
the serum with 3 vol of water, adjusting the pH to 7.6-7. 8 , cooling 
to 0 ° C, then adding cold (-10 to -20 C) 50 per cent ethanol 
to a concentration of 18-25 per cent. This results in the more or 
less complete precipitation of all of the antibody-active protein 
( 7 -globulins) with certain other proteins, notably ^-globulins. 
This 1st precipitate is designated A in subsequent discussion. 
Secondary precipitation steps, usually in the neighborhood of pH 5, 
are now used to remove the contaminating proteins. This 2d 
precipitation step is referred to as B. The supernatant 7 -globulins 
are now removed (as precipitate C) by adjustment of the pH to 
approximately 7.4 with NaHCOa and addition of ethanol to a 
concentration of 25 per cent. 

7-globulins.— It appears that all animal serums contain anti¬ 
body molecules of quite different electrophoretic mobility. In the 
case of the human and horse systems they resolve fairly well m 
electrophoresis in pH 8.6 diethylbarbiturate buffer to give 2 
families of antibody-active proteins which we have design^ed 
7 .- and 7 i-globulins in order of decreasing isoelectric point, ihe 
7 ;-globulin, which in alkaline buffers has the approximate electro¬ 
phoretic mobility of fibrinogen, has also been referred to as i- 
component, ft-globulin. etc. We prefer to call tins fraction |.- 
dobulin because of the usual association of the term 7 -globuhn 
with serum proteins possessing antibody activity- ® 

pig there appears to be a series of protein molecules in the y-glo^ 
ulfn fraction which, while showing marked electrophoretic he erc> 
geneity do not resolve into 2 components, ^evertheless it is 
possible to separate the latter antibody systems into fractions of 
ouite different average isoelectric points. Again we haw referied 
?o jLsfSons as 7 ^ and 7 .-globulins in order of decreasing 
isoelectric point. It is actually possible to separate a series 


fractions of varying electrophoretic properties from a given serum. 
Although our main emphasis in this discussion will be to present 
conditions for the preparation of the 7-globulin of highest iso¬ 
electric point (72-globulin), we shall discuss conditions for separa¬ 

tion of the 7 i-globulin fraction in certain instances. 

SUBFRACTIONATION. —Subfractionatioii of electrophoretically 
heterogeneous antibody protein is important, for preliminary 
investigations have disclosed that human and horse 72 - and 71 - 
globulin fractions show differences in antibody activity. Studies of 
the relationship of molecular mass, electrophoretic, solubility and 
other physical properties to various immunologic activities call 
for utilization of highly fractionated and carefully characterized 
antibody in which the molecules of any given fraction are as 
uniform as possible. 

The conditions for separating the 7 -globulins from the serum 
of various animals will be presented for each species. The reference 
after the name of each species indicates a publication describing 
fractionation of the discussed serum. 

Rabbit (13).—Since this species is utilized so extensively in im¬ 
munologic work, it will be used as an example of an animal whose 
serum antibody response appears to be largely in the electropho¬ 
retic fraction designated 72 -globulin. The details of fractionating 
100 ml of serum will be given. It will then suffice to present the 
fractionation of the other animal species in terms of conditions 
utilized at a given precipitation step or by reference to the analo¬ 
gous step used in the case of rabbit serum. 

One hundred ml of serum is diluted to 400 ml with distilled 
water and pH measured at 25 C and adjusted to 7.7 (= 0.1) 
with NaHCOs or dilute acetic acid if necessary. Sera collected by 
centrifugation of clotted blood that has stood in the cold overnight 
usually possess the desired pH. The diluted serum is cooled to the 
freezing point and a small amount of ice allowed to form. Now 
400 ml of 50 per cent ethanol, precooled to —10 to —20 C is 
added slowly by means of a siphon as the diluted serum is being 
stirred as vigorously as possible without causing foaming. Tem¬ 
perature oi the serum system is then brought dovm slowly to — 6 C 
as the ethanol concentration is increased and maintained near 
this point until all of the alcohol has been added. No ice crystals 

to solution after sufficient alcohol has been added 

to ouer the freezing point of the solution to -6 C For small 
scale operations such as this the diluted serum is placed in a glass 
beaker of sufficient size and immersed in an ice-salt bath or a 
cold ethanol solution to which small pieces of dry ice are added 
occasionally to lower the temperature to the desired level Lai^er 


volumes can be cooled by insertion of a refrigerated coil into 
the material being fractionated. Less efficient is the cooling by 
conduction of the solution container that obtains on standing iji a 
refrigerated room. 

Fifteen to 30 min after completion of the etlianol addition, 
precipitate A of crude 7 -globulins is remo^"ed by centrifuging undei' 
conditions that maintain a temperature between —5 and —7 C. 
The precipitate is suspended in 100 ml of water containing a few 


Precipitate B Precipitate C 

Fig. 2. —Descending electrophoretic j)atterns of rabbit serum and fractions in 
pH 8.6, 0.1 ionic strength, diethylbarbiturate buffer. 

ice crystals. Cold 0.05 M acetic acid is then added to pH 5.1. 
After stirring for 10 min the pH is redetermined and adjusted to 
5.1 if nece.ssary. Ionic strength is now about 0.002. The precipi¬ 
tate (B) which forms at this point is then removed by centrifuging 
near 0 ° C. The clear supernatant solution which contains the 72- 
globulin is adjusted to pH 7.4 (± 0 . 2 ) Avith cold 0.5M XaHCOs. 
An equal volume of cold ( — 10 to —20 C) 50 per cent ethanol is 
slowly added and the temperature of the solution lowered to -6 C. 
After addition of the ethanol is completed the 72 -globulin pre¬ 
cipitate (C) is removed by centrifuging at -5 to -7 C. One 
of the several procedures indicated previously is then instituted 
to bring the precipitated protein to conditions under which it 
can be stored without danger of deterioration. Electrophoretic 



patterns in pH 8 . 6 , 0.1 ionic strength, diethylbarbiturat.e buffer 
which indicate the course of the foregoing fractionation are sho^vp 

in Figure 2. 

The amount, of 7 .^-globulin obtained per unit, volume ot seium 
depends on the amount of this protein in the startiiig serum. 
This as is well known, is a function of the degree ot immune 
response of the animal(s) utilized. Conditions for recovery of the 
7 i-globulin of rabbit serum have not been determined. The major 
portion of these proteins is precipitated at pH 5.1 (preicipitate B). 

Goat, rat, guinea pig, dog (13).—Only the 72 -globuliii entity of 
these serums has been prepared. The fractionation pioceduies 
differ from those used for rabbit serum mainly in the conditions 
used for removal of precipitate B. are given in Table 1 
together with the conditions utilized for the rabbit. The super¬ 
natants to precipitate B in the case of the rat and dog contain 

TABLE 1. —Conditions for Precipitation B for Various Animaus 





























Guinea pig 





* Supernatants to precipitate B for these species contain 71-globulin that must be re¬ 
moved by an additional precipitation step before precipitation of 72-globulin. 

considerable amounts of 71 -globulin. This is removed in admixture 
with some 72 -globulin by a precipitation step at pH 5.6-6.0, n = 
0.005-0.01, at an ethanol concentration of 12 per cent. The 72 - 
globulin containing supernatant to this precipitation step in the case 
of rat serum is essentially free from 71 -globulin. All 72 -globulin prepa¬ 
rations from the dog contain 5-10 per cent of 71 -globulin. An 
extensive investigation of conditions for removal of this material 
has not been successful. Guinea pig serum could not be fractionated 
to give 72 -globulin preparations of high purity. Under the most 
optimal conditions found, a product which is approximately 80 
per cent 72 -globuliu is obtained. 

Man (7). Precipitate A is suspended in water under the usual 
conditions and adjusted to pH 5.1. Ethanol, 50 per cent, is added to 
a concentration of 10 per cent. The temperature should be -2 C, 
and the ionic strength is minimal and approximately 0.0015 at this 
point. The precipitate (B) is removed by centrifugation and 
discarded. The supernatant proteins are then precipitated at pH 
7.4 in the usual manner. This precipitate (C) consists almost 
entirely of 72- and 71 -globulin with a great preponderance of the 
iormer. It can be subfractionated by the following procedure. 


SUBFRACTIONATION. —The precipitate is suspended in water 
to give a concentration of approximately 2 per cent and dilute 
acetic acid is added to pH 5.1-5.3. The solution is stirred for 
approximately 15 min and any insoluble material is centrifuged 
off and discarded. The solution is then adjusted to pH 5.5 with 
Na 2 HP 04 and sufficient NaCl is added so that after the addition 
of 50 per cent ethanol to a concentration of 10-12 per cent an 
ionic strength of approximately 0.005 results. The precipitate 
formed under these conditions consists largely of 71 -globulins 
but should be refractionated under the above-described conditions 
at pH 5.5 one or more times to prepare material of relatively 
uniform composition. The supernatant to the first pH 5.5 pre¬ 
cipitate is adjusted to pH 7.4 with NaHCOs and 95 per cent ethanol 
is added to a concentration of 25 per cent. This precipitate con¬ 
sists largely of 72 -globulin and can be separated from about 10 
per cent of 71 -globulin by the following precipitation steps. The 
precipitate is suspended to give approximately 2 per cent protein 
concentration, the pH is adjusted to 5.1 with dilute acetic acid 
and sufficient NaCl is added so that after the addition of 50 per 
cent ethanol to a concentration of 20 per cent an ionic strength 
of 0.01 results. The small precipitate forming is removed and 
discarded, the supernatant adjusted to pH 7.4 with NaHCOs 
and 95 per cent ethanol added to a concentration of 25-30 per 
cent. The precipitate consists almost entirely of 72 -globulin. 

It is to be stressed that the re-precipitation steps, while re¬ 
sulting in yield decreases, are necessary to obtain material con¬ 
taining only small amounts of the other 7 -globulin fraction, i.e., 
if the isoelectric points of the molecules making up such material 
are to show very little overlap. Actually it is possible to separate 
a series of 7 -globulin fractions from human serum showing a gra¬ 
dation of physical and immunologic properties (5). 

Horse ( 6 ).—The commercial importance of horse antitoxin 
serum and the serum protein changes resulting during immuni¬ 
zation make this animal an especially interesting species to study 
from an immunologic standpoint. A particularly interesting feature 
of horse immune serum of the antitoxic type is the large amount 

of 7 i-globulin. , , 1 . • 

Precipitate A is removed in the usual manner and 
solution at pH 5.2 at minimal ionic strength (approx. 0.0015). 
Alcohol is added to 10 per cent concentration and the precipitate 
(B) formed is removed. The supernatant proteins are precipitated 
at pH 7 4 in the usual manner. This precipitate (C) consists of a 
mixture of 7 ^ and 7 ,-globulins, their ratio depending on the 
tog serum. Normal, veakly antitoxic and antibacterial serums 


show a higher amount of 72 -globulin. Strong antitoxic serum gives 
a precipitate C that contains larger amounts of 71 -globuhn. 

SUBFRACTIONATION. —To subfractionatc precipitate C, the ma¬ 
terial is brought into solution at approximately 1 per cent protein 
concentration at pH 5.4 and minimal ionic strength (approx. 


Precipitate A Supernatant to ppt. A 

Precipitate C 

Precipitate B 

Precipitate C -2 

Precipitate C-3 Precipitate C-U 

Fig. 3. —Descending electrophoretic patterns of horse serum and fractions in 
pH 8.6, 0.1 ionic strength, diethylbarbiturate buffer. 

0.0015). The insoluble material (precipitate C- 1 ) usually contains 
very little antibody and is discarded. The supernatant is adjusted 
to pH 5.8 and sufficient NaCl is added so that after addition of 
50 per cent ethanol to 10 per cent concentration an ionic strength 
of 0.005 is obtained. The precipitate (C- 2 ), consisting largely 


of 7 i-globuliii, is removed. The supernatant is brought to 20 per 
cent ethanol and precipitate C-3, consisting of a mixture of 72 - 
and 7 i-globulins, is removed. The supernatant is brought to pH 
7.4 and 95 per cent ethanol is added to a concentration of 25 per 
cent. The precipitate (C-4) consists largely of 72 -globulins. When 
fractionating serum containing large amounts of 71 -globulin it 
may be necessary to remove jjrecipitate C-3 at an ethanol con¬ 
centration of 25 per cent to prevent small amounts of this protein 
from being carried ovTr into the next precipitate. Conditions have 
not been found by which the major portion of the 72 - and 71 - 
globulins can be separated into single fractions, each relatively 
free from protems contained in the other fraction. Under the 
precipitating conditions employed, varying amounts of the 72 - 
globulins have solubiliHes close to the 71 -globulin, and vice versa. 
Individual serums show a great deal of \'ariation in the yields, 
and the physical character of the fractions separated, particularly 
precipitate C-3, cannot be predicted without previous fractiona¬ 
tion experience on a given immune serum. 

Electrophoretic patterns in pH 8 . 6 , 0.1 ionic strength, diethyl- 
barbiturate buffer showing the course of a typical horse serum 
fractionation are given in Figure 3. Unlike the freshly prepared 
7 -globulin fractions of other species, the horse proteins, when 
examined in the ultracentrifuge, show the presence of 10-20 per 
cent of material of approximately twice the molecular weight 
of the normal components ( 6 ). This material appears to be a 
product formed during fractionation, for examination of 7 - 
globulins recovered by electrophoretic separation shows only 

the normal component ot S 20 k) = 7N(14). 

Cow (10, 11).—The globulin fraction of this serum is represented 
by protein mole(*ules which, like those of the other animals, shov 
a wide isoelectric point range. They do not, howevei-, tend to sepa¬ 
rate into discrete components, during electrophoresis, as is true 
for the horse and man. Two fractions of different average iso¬ 
electric point possessing antibody activity may be separated. The 
conditions described below tend to give apiu-oximately equal 
amounts of the so-called 71 - and 72 -globulin fractions. 

Precipitate A is best removed at an alcohol concentration ot 
18 per cent. This material is separated from i 3 -globulins by pre¬ 
cipitation of the latter contaminants at pH 5.0, g - 0 . 01 . Ihe 
supernatant globulins are recovered in the usual manner as pre¬ 
cipitate C. This fraction is then divided into approximately equal 
portions bv adjusting a solution aiiproximately 1 pei- cent 
to pH 5 8 u = 0.01 and ethanol 10 per cent. Ihe supernatant to 
tho^procipitatP (C-1) wliicli forms is ridjuslod to pf l.,.4 arrd otiiariol 



25 per cent. The higher average isoelectric point fraction (pre¬ 
cipitate C-2) designated as 72 -globulins is then removed. 

The electrophoretic diagrams in pH 8 . 6 , O.I ionic strengt 1, 
diethylbarbiturate buffer which follow the course of this fractiona,- 
tion are shown in Figure 4. The 72- niid 71 -globulm fractions each 
contain considerable amounts of protein common to the other 

Precipitate B 

Precipitate C 

Precipitate C-1 Precipitate C-2 

IiG. 4.—Descending electrophoretic patterns of cow serum and fractions in pH 
8.b, 0.1 ionic strength, diethylbarbiturate buffer. 

fraction; i.e., 
utilization of 
to increas{‘ 01 
Jioint. An iiu 

they show a pronounced overlap in isoelectric point 
The average isoelectric point and electrophoretic 
oi either 7 -globulin fraction may be altered by 
conditions at, the C -1 precipitation step which tend 
• decrease the amount of protein precipitated at this 
•rease in the amount of 71 -globulin fraction precipi- 


tated, which may be mediated by increasing the ethanol concentra¬ 
tion above 10 per cent will result in a small yield of 72 -globulin 
of higher isoelectric point and decreased electrophoretic hetero¬ 
geneity. Conversely, a 71 -globuhn fraction of lower average iso¬ 
electric point and heterogeneity may be recovered by decreasing 
the amount of ethanol used at the C-1 precipitation step. 

Pig (I).—The 7 -globulin fraction of the pig resembles that of 
the cow as regards the electrophoretic heterogeneity of the protein 
fractions that are obtained. In this case the 72 - and 71 -globulin 
fractions separated show a greater heterogeneity than is seen for 
the bovine systems. It is also more difficult to remove the /3- 
globulins from the 7 -globulin fraction. 

Precipitate A is removed in the usual manner but at an ethanol 
concentration of 20 per cent. It is suspended to give a 1 per cent 
protein solution and pH is adjusted to 5.2. After stirring for 30 
min, precipitate B is removed. The crude 7 -globulins in the super¬ 
natant are precipitated with 25 per cent ethanol at pH 7.4. 

Precipitate C is suspended in water to give a 1 per cent protein 

solution and adjusted to pH 5.6. Sufficient NaCl is added to give 
an ionic strength between 0.005 and 0.01 when 50 per cent ethanol 
has been added to a concentration of 15 per cent. The precipitate 
(C-1) consists of crude 71 -globulins with 5-10 percent jS-globulins 
as impurities. It is further fractionated by suspending in water 
and removing a precipitate at pH 5.2, ionic strength 0.005 and 
ethanol concentration 5 per cent which consists largely of /3- 
globulins. The supernatant 71 -globulins are recovered by precipita¬ 
tion at an ethanol concentration of 25 per cent at pH 7.4. 

The 72 -globulins in the supernatant to precipitate C-1 are re¬ 
covered at an ethanol concentration of 25 per cent at pH 7.4 and 

ionic strength approximately 0.01. 

As in the case of the bovine globulins, the amounts and elec¬ 
trophoretic properties of the 72 - and 71 -globulin fractions may 
be altered by modification of the conditions giving rise to pre- 

clfcLn ( 13 ) —The serum of this species is characterized on elec¬ 
trophoretic study in pH 8.6 diethylbarbiturate buffer by absence of a 
component corresponding in mobility to the mammalian 72-globu in. 
Two distinct 7-globulin fractions which we have also designate 72- 
am) T.-globulini are present, l>ut t)rey Irave considerably h.gFier 
mobilities than the proteins of similar biologic function have m 
higher animals. In addition, there appear to be protem(s) Pok¬ 
ing the mobility of a-globulins in some antisera which can add oi 
^aiitige'-antilLly complexes even though they are unable to orm 
precipfates with antigen in the absence of chicken y-globuhn (8). 


Precipitate A is taken off under the usual conditions at pH 7.7 
and 25 per cent ethanol. This material is then fractionated at pH 
5.0, n = O.OI and ethanol 10 per cent to give precipitate B. The 
7 -globulin-containing supernatant is then precipitated between 
pH 7.0 and 7.4 at an ethanol concentration of 25 per cent. The 
7 -globulins recovered at this step give asymmetrical electrophoretic 
patterns in pH 8.6 buffer and appear to be a mixture of 72 - and 
7 i-globulins. They differ from mammalian 7 -globulins in giving a 
sedimentation constant of 7.8-8.0 S. The 7 -globulin fractions also 
often contain varying amounts of serum conalbumin (transferri¬ 
tin) which may be readily recognized on ultracentrifugal analysis 
or by determining the iron-binding capacity of the 7 -globulin 
preparation. The conalbumin can be removed b.y repeated precipi¬ 
tation of precipitate C at m = 0 . 01 , pH 7.0-7.4 and an ethanol 
concentration between 20 and 25 per cent. 

The nature of the a-globulin component which can enter into 
the antigen-antibody reaction is poorly understood. It is soluble 
at the conditions used to remove precipitate C. The differences 
in the appearance of specific precipitates formed by addition of 
antigen to whole serum and of 7 -globulin fractions suggests that 
the a-globulin factor is lipoprotein in character. The ability of a 
factor in the a-globulin to add on to an antigen- 7 -globulin system 
in this species to give markedly increased amounts of specific 
precipitate is an example of complications that can be avoided 
and controlled in immunologic work when one utilizes antibody 
preparations of the 7 -globulins rather than whole serum. 

1 . 

2 . 




6 . 


Camniarat-a, P. S., and Deutsch, H. F.: Biophysical studies of blood 
plasma proteins: XIV. Separation of gamma globulins from normal hog 
serum, Arch. Biochem. 25: 354, 1950. 

Cann, J. R., el al.: Ultracentrifuge studies of y-globulins prepared by 
electrophoresis convection. Science 114: 50, 1951. 

Cohn, E. J., el al.: A system for separation into fractions of the protein 
and lipoprotein components of biological tissues and fluids J Am 
Chem. Soc. 68: 459, 1946. ' 

Cohn, E. J el al.: A system for the separation of the components of 
human blood: Quantitative procedures for the separation of the pro- 
tern components of human plasma, J. Am. Chem. Soc. 72: 465 1950 

Cohn, M.; Deutsch H. R and Wetter, L. R.: Biophysical studiefof 
blood plasma proteins: XIII. Analysis of immunological heterogeneity 
of human gamma globulin fractions, J. Immunol. 64: 381 1950^ ^ 

eutsch, II. F and Nichol, J. C.: Biophysical studies of blood plasma 

TBbrcL -™">. 

Deutsch, H. F.; Alberty, R. A., and Costing, L. J.: Biophysical studies of 

Hn from'^nT*' Reparation and purification of a new globu¬ 
lin from normal human plasma, J. Biol. Chem. 165-21 1946 ^ 


8. Deutsch, H. F.; Nichol, J. C., and Cohn, M.; Biophysical studies of 

blood plasma proteins: XI. Immunological and electrophoretic studies 
of immune chicken serum, J. Immunol. 63: 195, 1949. 

9. Ileidelberger, M.; Krueger, R. C., and Deutsch, H. F.: Antiproteins in 

horse sera: VII. Fractionation by alcohol of antibodies to the globulin 
of rabbit serum. Arch. Biochem. 34: 146, 1951. 

10. Hess, E. L., and Deutsch, H. F.: Biophysical studies of blood plasma pro¬ 

teins: VIII. Separation and properties of the gamma-globulins of the 
sera of normal cows, J. Am. Chem. Soc. 70: 84, 1948. 

11. Hess, E. L., and Deutsch, H. F.: Biophysical studies of blood plasma pro¬ 

teins: IX. Separation and properties of the immune globulins of the 
sera of hyperimmunized cows, J. Am. Chem. Soc. 71: 1376, 1949. 

12. Krueger, R. C., and Heidelberger, M.: Effect of the removal of lipis on 

specific precipitation, J. Exper. Med. 92: 383, 1950. 

13. Nichol, J, C., and Deutsch, H. F.: Biophysical studies of blood plasma 

proteins: VII. Separation of 7-globulin from the sera of various animals, 
J. Am. Chem. Soc. 70: 80, 1948. 

14. Savat, R. S.: “Physical Studies of the Antibody Active Proteins of Vari¬ 

ous Animals.” Bachelor of Science thesis. University of Wisconsin, 

15. Smith, E. L., and Brown, D. M.: The sedimentation behavior of bovine 

and equine immune proteins, J. Biol. Chem. 183: 241, 1950. 


A. Reaction in Liquid Media 



We will discuss a quantitative method for following the reaction 
which results in the formation of a specific precipitate between 
a soluble antigen and an antibody. This insoluble precipitate is 
freed from nonreactive substances by centrifugation and washing 
with buffer or saline and the quantity of nitrogen precipitated is 
determined by a micro Kjeldahl method. 

The particular variation which we prefer has a number of dis¬ 
tinct advantages over other techniques. However, it is one of a 
dozen procedures and is presented only because it has not been 
discussed in full elsewhere.* It can be changed and simplified in 
several ways. The problems involved in carrying out this procedure 
are: (1) to avoid nonspecific precipitation; (2) to have the system 
in equilibrium, and (3) to separate completely the specific pre¬ 
cipitates from the supernatant fluids. 

Certain general features deserve mention. It is preferable, but 
not always essential, to fractionate the antiserum for the antibody- 
containing 7 -globulin. This eliminates complement t (see Deutsch, 
this volume), certain lipids which separate on standing and, very 
often, nonspecific precipitation. The choices of conditions for the 
antigen-antibody reaction must be compatible with the stability 
of each reactant. For most cases, a pH close to neutrality and 
a salt concentration close to isotonicity are suitable. If the antigen 
is a protease, its possible action on the antibody must be con¬ 
sidered and conditions should be chosen such that the enzyme has 
no significant activity. 

There are several ways to begin the study of an unknown system 
and to evaluate these factors. The following procedures are suffi- 

* <:ertaiii equipment discussed here is not available, see Rabat and Mayer (11) 
and Loiseleur (12), for detailed description of other techniques, 
n u n addition of complement to the specific precipitates will not spe- 

^ results. However, m work with very small amounts of antibody, the 

problem is iinportant and complement must be removed; for the technique, see (1). 
Another useful means of eliminating complement is to allow the sera to remain in 
the refrigerator until complement activity disappears. 



ciently controlled and reliable, but they can be carried out in many 


a) Conical tuhesX (Fig. 1).—These must be made of the highest 
equality pyrex glass to withstand repeated heating and centrifu¬ 
gation. They can be obtained calibrated to facilitate colorimet¬ 
ric reactions for determination of total protein precipitated. All 

Fig. 1. —Conical tube and stirring apparatus for resuspending specific precipi¬ 
tates. Length of tube, 15 cm; width, 1.8 cm. 

operations up to the actual nitrogen determination are carried out 
in the same tube. 

h) Technique .—Before carrying out the reaction, thoroughly 
centrifuge antigen and antibody at the same speed that will be 
used in later centrifugations. Choose a serum diluted to contain 
about 150 jug antibody in 1 ml, as determined by preliminary tests 
(p. 311). To a series of conical tubes add varying amounts of 
isotonic buffer (or 0.15M NaCl) in volumes calculated to give a 
final fixed volume between 2 and 4 ml. Within the limits of solu¬ 
bility considerations, the precipitin reaction is volume-independent. 
Increasing amounts of antigen (starting at 2 ^g N for protein 

t Made on order by E. Machlett & Son, 218 E. 23d St., New York 10, and by 
Delplanque, 212 me Saint Jacques, Paris, France. 



and 10 Mg for polysaccharides) are then added, followed by the 
1 0 ml fixed volumes of antibody solution. The latter should be 
added rapidly and the tube contents mixed immediately there¬ 
after by rolling the tubes rapidly between the hands. This is done 
to avoid the Danysz effect (2). 

One of the greatest hazards in carrying out the quantitative precipitin 
reaction is the possibility of nonspecific precipitation. This is particularly 
marked with crude bacterial extracts, proteins of very high nmlecular 
weight and proteins with isoelectric points at the extremes of pH. Non¬ 
specific precipitation can sometimes be avoided by choice of conditions of 
precipitation or by fractionation of antisera. In certain cases it occurs only 
at protein concentrations much higher than those used in the specific re- 
action.§ As a check on nonspecific precipitation, controls must be set 
up, including the antiserum alone plus buffer, antigen plus normal serum 
obtained before immunization, and antigen plus unrelated antiserum. No 
precipitation should occur in these controls. If nonspecific precipitation 
cannot be avoided, a control with normal serum can be set up for each 
mixture, and this value subtracted from that obtained with antiserum. 
Results under these conditions are at best rough approximations. 

The various mixtures should be set up in duplicate. In case of 
shortage of antigen or antibody, it is better to have double the 
number of different mixtures than fewer in duplicate. 

Certain preliminary tests must be made to determine correct 
conditions for studying the reaction, in particular the effect of 
incubation and washing. Three different amounts of antigen, 
chosen on the basis of preliminary tests (p. 311), should be added 
to constant volumes of antiserum, such that the resultant systems 
are (a) in antibody excess, (h) near the equivalance point and (c) 
in antigen excess. Each of these should be prepared in octuplicate. 
In general, 4 incubation conditions can be tried, with 2 tubes at 
each point: (1) 4 C for 2 days; (2) 4 C for 5 days; (3) room tem¬ 
perature until precipitation occurs (2-3 hr) and 2-3 days at 4 C; 
(4) 40 C until precipitation occurs (1-2 hr) and 2-3 days at 4 C. 

The tubes are centrifuged after the various incubations; 1 tube 
of each pair is washed 3 times and the other 5 times (see B below) 
and N content of the precipitates determined. Thus in 1 prelimi¬ 
nary experiment the effect of incubation and of washing is evaluated. 
In most cases the precipitate is sufficiently insoluble so that no 
significant difference between 3 and 5 times washing will be notice- 

5 With certain substances such as ricin which react nonspecifically with serum 
proteins, estimation of anybody in specific precipitates is not reliable. Although only 
high concentrations of ricin alone will react with normal serum, specific precipitates 
^ antibody, even with concentrations below that at which ricin Visibly 
serum, appear to contain both antibody and nonspecific proteiS 

^ *1^ K would be interesting to know whether the isolated antibody-con- 
tainmg globulin also reacted nonspecifically. annuouy con 


able. If the specific precipitate is appreciably soluble, washing 
must be minimal, and correction for solubility can be made. 

Many factors are involved in the attainment of complete pre¬ 
cipitation. It is more difficult to obtain reproducible results in the 
region of antigen than in the region of antibody excess. In cases 
of difficulty, incubation at 4 C for long periods (7-14 days) has 
given reproducible results. Often, however, for reactions involving 
more than 100 /xg antibody N, maximal precipitation occurs after 
2 days at 4 C and further incubation has no effect. These factors 
must be tested in each new system, since they may vary consider¬ 
ably. With certain antigen-antibody reactions, the same results 
will be obtained under any of these conditions, in which case 
it is preferable to incubate at 4 C for 2 days. In the presence of 
viscous solutions, e.g., bacterial extracts or high protein concentra¬ 
tions, preliminary incubation at high temperatures (37-45 C) is 
sometimes necessary before storage at 4 C in order to obtain equi¬ 


a) Apparatus. —1. Centrifuge. We use the International Re¬ 
frigerated Centrifuge model PR-1 with no. 269 Mitat head. Eight 
tubes can be spun at 1 time at 3,000 rpm. Although the tubes 
fit loosely into the no. 320 Atab cups, the only necessary precaution 
is to turn the pouring spouts to the outside, since breakage might 
otherwise occur when the long tube, spinning horizontally, touches 
the head. 

2. Stirrer." The specific precipitates can be resuspended in the 
washing solution with a small glass stirring rod drawn out to a 
point. However, in work with large numbers of samples or with 
voluminous precipitates, a power-driven stirrer is preferable. A 
fine spiral propeller of stainless steel (Fig. 1) turning at about 
1,000 rpm is very efficient. It should be well guided so that it turns 

perfectly centered. 

Comment hy E. A. Rabat 

Experience in our laboratory has led us always to avoid a procedure in 
which a stirrer or any other object was placed in the tube for wa.shing 
specific precipitates. After the tubes have drained, the precipitates can be 
broken up by tapping the tubes vigorously; 0 5 ml chilled saline is add^ 
and the precipitates are uniformly suspended in this small volume. The 
additional volume of wash liquid is then added and the precipitates aie 




dispersal by gentle twirling. By this procalure, the need for 
introducing a stirrer in the tiil)C is avoided, 

h) Technique.—nw. (iilx's are centrifuged at O'’ C for I hr at 
3 000 rpni for the 1st separation of supernatants, which should be 
absolutely clear. Even if a given tube seems to have no precipitate, 
it should be centrifuged and treated as the others because certain 
precipitates are very hard to see in the presence of serum. In some 
cases these supernatants remain turbid, but often they may be 
clarified by centrifugation at higdier speeds (10,000 rpm in a Servall 
angle centrifuge), or in certain cases, if fractionated antisera 
are used, turbidity will be avoided. If a refrigerator room is not 
available, the tubes are kept cold in an ice bath. The supernatant 
is carefully poured off and saved, and the conical tube is allowed 
to drain upside down on blotting paper. Four ml of cold buffer 
or saline is added so that it washes down the tube wall. The pre¬ 
cipitate is resuspended and finely divided with the mechanical 
stirrer or simple glass rod, and the stirrer washed down with an¬ 
other 1 ml of wash solution. The tubes are then centrifuged, and 
the operation is repeated 3 times. The recentrifugations can be 
made for 15 min periods. 


a) Materials. —1. Digestion mixture: Various solutions have 
been described—all quite good. The simplest is made by adding 
cautiously 20 ml of a saturated solution of aqueous copper sulfate 
to 1 1 of concentrated sulfuric acid (Merck C.P., “low nitrogen”). 
The excess anhydrous copper sulfate settles out after several days, 
and the supernatant is used for the digestion process, 

2. Potassium sulfate: Special low nitrogen analytical reagent 

3. Digestion rack: One of the most convenient is an electrically 
heated sand bath with temperature controlled by a rheostat in 
the external electric circuit. Figure 2 shows that used at Pasteur 
Institute. The heating elements (1,500 w) are covered with sand, 
and the conical tubes are placed through a support at a 45° angle 
into the .sand so that only the conical part is below the surface. 

4. ‘Distillation apparatus: We prefer that made by Scientific 
Glass Apparatus Company, Bloomfield, N. J., because it is compact 
and constructed with all-glass connections. 

5. Titiation. Although most laboratories use indicator titration, 
electrometric methods are more precise and reliable and much less 
fatiguing. We have used the quinhydrone electrode successfully 
but are now using the Beckman pH meter model G with shielded 
glass calomel outside electrodes no. 11-305-70 and 8970-90 The 


sample is mixed with a magnetic stirrer, which does not affect these 
electrodes. Titration is carried out with a Scholander buret (3) 
filled with acid, so that 100 divisions of the micrometer equals 

s ✓axnvx\x\v\x\v\.\\vn\.xn\x\wx 

U Reststdnce\/^ 


Rsbestos plates^JR^ L- 

i«E- 500mm - 

I ' 

20 Mg N. This standard acid is prepared by dilution of a weighed 

amount of constant boiling HCL (4). • • u + /-i^^+ar 

h) Technique.— The quantity of specific precipitate is best deter¬ 
mined by N estimation. For initial studies unknown pjtems 
To otW method is satisfactory. When the reaction is well defined 
and when the antigen or antiserum is to be used tor assay purposes, 

the rapid, simple and often more sensitive absorption and density 

determinations are of great value (5-10). u . u 

Five-tenths ml of the digestion mixture is added to each tube, 
plus 250 mg of potassium sulfate. The latter is conveniently added 
uith a little scoop made out of brass rod which will just hold 250 mg. 
The tubes are transferred to the digestion rack. No boiling stones 
are used, and the acid, without ebullition, refluxes gently about 
halfway up the tubes. In general, after about 3-4 hr of reflux, 
digestion is complete. The time of complete reaction should be 
checked when working with a new system. 

The conical tube contents are transferred to the distillation 
apparatus by repeated washing with a total of 30 ml water. Then 
3 ml of a saturated NaOH solution is added and distillation begun 
into 3 ml of a saturated solution of boric acid in a 50 ml beaker 
scratched to indicate volumes equal to 20 and 25 ml with a diamond 
point pencil. The exit of the distillation apparatus is placed be¬ 
low the surface of the boric acid solution, and after the volume 
of distillate reaches the 20 ml mark the beaker is lowered so that 
the opening is out of the liquid and distillation continued to the 
25 ml mark in order to rinse the condenser. Distillation is stopped. 
The tip of the refrigerant column is rinsed with 2 ml water and 
the sample is ready for titration. For titration, the acid is 
added to pH 5.0 from the Scholander buret whose tip is placed 
just below the surface of the liquid. This determination is quite 
accurate with N concentrations between 10 and 150 /xg, which is a 
sufficient and convenient range for immunochemical work of the 
type discussed here. 


For interpretation of results it is essential to know whether 
there is excess of antigen or antibody in the supernatants. This is 
carried out qualitatively as described below. 

On the basis of the quantitative precipitin reactions, a general 
idea of the sensitivity of the test for excess antigen or antibody 
can be obtained. However, it is important to know the sensitivity 
with precision, and this is done by adding to fixed volumes of 
antiserum (containing about 100 /xg antibody N) increasing 
amounts of antigen, 0.1, 0.5, 1.0, 2.0, 5.0 ng. These mixtures are 
incubated as for the original reaction and centrifuged. A very 
small precipitate is most easily detected as a centrifuged pellet. 
The concentration of antigen which is just detectable as precipitate 
IS the limit of our tests. This value is generally between 0.5 and 
1.0 Mg antigen. The test for sensitivity of antibody detection is 
carried out by adding dilutions of antiserum corresponding to 


between 3 and 10 ng antibody N to 1 and 5 mS antigen. The mixture 
in which the least amount of antibody is detectable decides the 
amount of antigen to be used to test for antibody excess. Since 
detection of antigen excess is more sensitive than that of antibody 
excess, V 2 of the supernatant is used for the former and for the 
latter. The method of analysis depends on the results of the pre¬ 
liminary tests; one uses the conditions which give maximum 

In the case of antigens which have measurable specific properties, 
i.e., enzyme, virus, hormone or chemical activity, distinct grouping 
as heme, distinct absorption as nucleic acid, such properties should 
be assayed in the supernatants, as interpretation of results is 
greatly simplified with this additional information. 


This discussion will be devoted to an analysis of several mode 
systems which can serve as a guide for interpretation of the more 
complex results often observed. For details on background and 
derivation of the relationships to be discussed, the reader is re¬ 
ferred to Rabat and Mayer’s book (11). 

Precipitation of the antigen-antibody complex is described by 
2 different behaviors—the precipitin and the flocculation reaction, 
represented by the graphs in Figures 3 and 7. The precipitin re¬ 
action is characterized by precipitation in the region of excess 
antibody and a narrow equivalence zone, where neither antigen 
nor antibody is detectable in the supernatant. The flocculation 
reaction* is characterized by a zone of complete solubility 
complex in excess antibody and a broad equivalence zone. In both 
systems the antigen-antibody complex is soluble in excess antigen. 
Each svstem will be discussed separately. 

iii auju-- . , 

held true for human 72-globiuin 
antibody response was excellen 
was found. 


cipitatcd (TN) minus antigen nitrogen added (X) equals antibody 
nitrogen precipitated (AbN). t 

When AbN is plotted against antigen added, the curve which 
describes this reaction is parabolic up to the point where all of 
the antibody is precipitated. Heidelberger and Kendall (13) have 

Fm. 3.—Curves of the precipitin reaction. Curve A, described by equation 
(3) (TN vs. X), represents the reaction of a pure protein. Curve A' is the same 
reaction descnbed by equation (1) (AbN vs. X). Curves B and B' are the same 
reaction except that the antigen is 25 per cent pure in the presence of inert 

protein. Arrows represent equivalence points. Extensions of curve A ( -) 

illustrate behavior of polysaccharides, complex systems, etc. (see text). 

derived the following simple relationship which satisfactorily 
describes the precipitin reaction in the region of excess antibody. 

AbN = 


where AbN = antibody nitrogen precipitated; X = antigen pre¬ 
cipitated, J in terms of nitrogen for proteins ;§ E = ratio ^ 

- antigen N 

o ^ polysaccharides which contain no nitrogen TM — a km r i 

P-iPita W Jn” ally 

Equation (1) woul^al^’dlj^scri'L^h^reactlon'^i^^^ excess zone, 

curve would be elongated along the abscissrhv T ® N added, but the 

centage of X in the mixture aosci.ssa by an amount depending on the per- 

^bN and .Y are 


at the equivalence point; A = total antibody N in the unit volume 
of serum. If we divide through by X, we have 

^ antibody “N precipitated _ ^ v 

antigen N precipitated ~ ~ X ^ 

The plot of r against X then gives a straight line with slope - — 


and intercept at the ordinate of 2R. 

In actual practice, it is TN which is determined. As we mentioned 
above, up to the equivalence point 

TN = AbN + X 

and from equation (1) 

TN = (2i2 + 1)X - (3) 

Let us now consider a hypothetical single antigen-antibody 
reaction in which the antigen is 100 per cent pure. R and A have 
arbitrarily been assigned the values of 4.0 and 1.0 mg/N unit 
antiserum, respectively. Curve A, Figure 3, represents this single 
antigen-antibody reaction and the equations describing it are 

TN" = 9X - 16X* 

AbN = SX - 16A* 

r = 8 - 16X 

The discussion has, until now, been concerned with behavior 
of the precipitin reaction up to the equivalence point. We have 
emphasized the equivalence point for the precipitin system in 
order to contrast it with the equivalence zone in the flocculation 
system. The equivalence region in the precipitin reaction is very 
narrow,* and in systems where the antigen is biologically active 
and can be detected in trace amounts, it can generally be shown 
that this zone does not extend over more than 5 per cent on either 
side of the point of maximal antibody precipitated (14). The 
equivalence zone found in the flocculation system often extends 

" It should be pointed out that the maximum for the curve of TN vs. X is dis- 
olaced relative to that of AbN; the maximum of the former occurs m the antigen 
excess region; aU antibody is precipitated before this point. The difference ^tween 
the X maxima necessary to bring equations (1) and (3) to TN maximum depends 
in theTvXe. For example, if 12 = 1. the difference between the X m_axima of 
eouations (1) and (3) is 33 per cent. If 72 = 6 it is 9 per cent, and for - 10 it is 
5 ner cent The increase in TN between the point of disappearance of antitedy from 
the suDeriiatant and the appearance of excess antigen is ^so a function of 72 value, 
and for 72 = 1 the increase is 10 per cent, and for « = 4 it is 1 per cent. Smce most 
o+i.rUoH havfi 72 values of 4 and larger, this increase in N precipitated is not 
proteins . However the difference in the 2 X maxima is often observed 

testm* supernatant, i. very 

*''i*pSionged immunization seem, to broaden this considerably (16). 


over 100 per cent on either side of the point of optimal flocculation 

Analysis of the antigen excess zone is much more difficult. 
Behavior of the precipitin reaction in this region has not been 
extensively investigated because (1) equilibrium for precipitation 
of the complex is hard to attain, (2) no theoretical relationship 
has been devised to describe it and (3) its quantitative signifi¬ 
cance is obscure. 

Recent investigations (14, 17, 18, 21, 23-25), however, show 
that this region is extremely useful in the detection of impurities. 
In general, it may be said that for native globular proteins, the 
excess of antigen necessary for complete solubilization of the anti¬ 
gen-antibody complex is usually between 2 and 3 times that neces¬ 
sary to reach the equivalence point. For curve A, Figure 3, 0.5 mg 
of added N is required to bring the curve to zero, whereas 0.25 mg 
is needed at the equivalence point, so that the value here is 2. 

One will find systems which approach this description of the 
precipitin reaction more or less closely. Many known precipitin 
reactions have been shown to fit the equations up to the equiv¬ 
alence point, t One of the most interesting, described by Adams 
(19), is the tyrosinase-antityrosinase reaction in rabbit and human 
sera, in which equation (1) was shown to hold remarkably well 
even in antigen excess. These observations were extended and 
confirmed with 2 antigens, diphtheria toxin (14) and jS-galacto- 
sidase of Escherichia coli (21), in which the special nature of 
the systems allowed preparation of an antiserum which reacted 
with a single antigen species. J 

High molecular weight polysaccharides (26, 27), artificially 
diazotized antigens (28), high molecular weight proteins (29, 38), 
denatured and relatively insoluble proteins (31) exhibit a slightly 
different behavior. Up to the equivalence point their reactions 
generally obey the relationships already discussed. However, 
the antigen-antibody complex is quite insoluble in the antigen 
excess, and the curves representing the reaction extend out into 

straight line instead of a parabola. This reaction^ int?rpJted to ^ 

coSefis%Scted'&us™^^^^^ 'he antigen-antibody 

toxins appSr fn^tL'Sturi fluiSo^T wi^h®''”'^-concentration, 
proteins L higLr iron concentration proteins, called P 

P proteins appear. Absorption of any antitoxin with°P ^ut the same 

antibody (14). The same fechSiJie was S ^studv ^ single 

enzyme of E. coli. Absorption of antisera with adaptive 

contain negligible amounts of ^-galactosidase resultsln^an 

a single antigen species (21). antibody that reacts with 




the antigen excess region in a manner shown by the dotted line 
continuations of curve A Figure 3. The insolubility in antigen 
excess is probably due to the presence of repeated combining 
groups on a very large molecule in the first 3 categories of sub¬ 
stances mentioned, while in the last 2 , insolubility of the antigen 
itself might be a factor in the insolubility of the antigen-antibody 
complex in an excess of antigen. 

The single antigen-antibody reaction discussed thus far can be 
obtained by using an antiserum which reacts with a single antigenic 
species or by using, as precipitinogen, an immunologically pure 
protein. Let us assume that the antiserum used to give curve A, 
(Fig. 3) reacted with a single antigen species, and as precipitinogen 
we used a preparation 25 per cent pure. The resulting curve would 
be identical to curve A when the abscissae are correctly reduced. 
Curve B illustrates this point. It can be seen that 4 times as much 
total antigen must be added to precipitate the same amount of 
total N in the case of B as in A therefore preparation B contains 
V 4 as much of the given antigen as does A. This is the principle of 
the immunologic assay for a given antigenic species. One must be 
certain that the antiserum does not contain a mixture of unrelated 
antibodies (discussed more fully later). 

The single antigen-antibody reaction is often unrealized and one 
is faced with the more complex behaviors due to simultaneous 
reactions between several antigens and antibodies. We will present 
as a typical system an imaginary case analogous to some that 
have actually been found. The analysis depends on the following 
assumptions: ( 1 ) that each reaction takes place independently of 
the other,§ and (2) that the equation of Heidelberger and Kendall 
(13) is applicable to each. 

Let us consider 2 precipitinogens whose reactions with their 
respective antibodies can be represented by the equations already 
discussed. Equation (3) (TN) will be used as a 1st approximation 
and it will be assumed that the equation applies in antigen excess. 
Consider now 2 antigens and antibodies {1 and 2) with the follow¬ 

ing constants 


Ai max 




20 Xi 

21 X, 

100 X\ 
100 X\ 


A* max 




10 X 2 

11 Xi 

250 Al 
250 XI 


Suppose that a given protein preparation contained 90 per cent 
Ah and 10 per cent Ah. Figure 4 represents the reactions of this 
system. At 0.11 mg of added antigen N, component 1 is at its 

eciuivalence point, wliereas there is 0.01 mg antibody N to com 

ponent ^ stdl unprecipitated in tlm supernatant. Tests of X 

supernatant at this point will show antibody excess A further 
addition of total antigen will result in an antigen excess of com¬ 
ponent 1 and antibody excess to component Thus the clearest 


©vidcncc for more than 1 reacting substance is the presence of 
both antigen and antibody in any given supernatant." We owe 
the precise statements of this condition for immunologic homo¬ 
geneity to Kendall (35), Culbertson (36) and Weil (37). 

Presence of an equiv'alence point would be a sufficient criterion 
of immunologic purity if the precipitin reaction as a test for antigen 
and antibody excess were sensitive enough. In fact, in the rather 
obvious example giv^en here, one might have missed the zone of 
simultaneous excess of both antibody and antigen because the 
test for excess of antibody by the precipitin reaction will detect 
at best about 3-5 Mg N in 2-5 ml. If component 1 were 80, not 90, 
per cent of the total and component 2 were 10 per cent, the rest 
being inert, the zone of excess antigen and antibody would not 
have been detected (see Fig. 4). At the equivalence point of com¬ 
ponent 1 , only 1 Mg antibody N against component 2 would remain, 
and we would say that the system ot 2 components had an equiv¬ 
alence zone. This being our only criterion, the 2 component re¬ 
actions would be called a single one. Now let us look at the antigen 
excess zone. The summation curve drops sharply, then levels off 
and trails out into far antigen excess. This does not in itself mean 
that the system is inhomogeneous, but it acts as a warning to look 
further. Examples of this type of complex curve were given by 
certain preparations of diphtheria toxin (14), ovalbumin and 
conalbumin (17), jS-lactoglobulin (23), 7 -globulin (18), ^-galacto- 
sidase ( 21 ). 

To test the system, another mixture of antigens is used as pre¬ 
cipitinogen, e.g., the starting material from which component 1 
Avas purified. For demonstration, suppose that the starting system 
consisted of 20 per cent component 1, 70 per cent component 2 and 
10 per cent inert material (see Fig. 4). At 0.028 Mg antigen N added, 
component 2 is at its equivalence point. Between this point and 
0.50 Mg antigen N added, excess of both antigen and antibody would 
be easily detectable in the supernatant, whereas with the purified 
antigen it would not be observed. Examples of investigations 
using this technique are those on conalbumin (17) and / 3 -lacto- 
globulin (23). Scherp (40) devised a method for resolving the re¬ 
action due to each component in that mixture and for determining 
the antibody nitrogen to each component. 

will only be mentioned neie. no e > , , , addition supernatants can also 

betoM by thrSodf thj:, eltectmc . considerable increase in the reliability ot 

these tests. 



In summary, then, the testing of a given purified protein anti¬ 
gen A should be done in the following manner. 

1) Anti-A sera should be reacted with (a) antigen A itself, (6) the 
crude starting mixture from which A was isolated and (c) fractions 
which contain higher concentrations of probable impurities. 

2) Antigen A should be reacted with antisera prepared against 
materials likely to contain impurities in A, such as the crude 
starting material or fractions of it. For example, anti-whole egg 
white sera, anti-whole serum sera, anti-milk whey sera and anti- 
allantoic fluid have been used to test purified ovalbumin (17), 
7 -globulin (18), )8-lactoglobulin (23) and encephalomyelitis virus 
(41), respectively. 

3) In cases of doubt due to very weak reactions, the antisera can 
be concentrated by fractionation and restudied (for an example see 
(17)). A single antigen-antibody reaction will show an equivalence 
zone, and in the case of native globular proteins of molecular 
Aveights up to about 200,000 the antigen-antibody complex will 
be readily soluble in excess antigen. The reverse, i.e., a curve 
Avhich drops sharply in antigen excess, is not by itself evidence 
of a single antigen-antibody reaction, nor is the trailing out of the 
curve into far antigen excess necessarily evidence of reactions 
of impurities. All of the previously discussed factors must be 
considered. The antigen excess region of the curve should be used 
only as a guide. All of the curves obtained under condition (I) 
should be superimposable over the entire region of precipitation. 

If an antiserum is to be used to assay for a given antigen in a 
complex mixture, it must be demonstrated either that the anti¬ 
body preparation will react with only a single antigenic species 
or that the reactions of impurities and antibodies to them are not 
significant in the range of values studied. It is quite possible that 
antibody directed against a trace of impurity, too small to detect 
in the purified antigen, would become very significant in the 
analysis of crude mixtures containing a greater concentration of 

very important that condition (I) be stud- 
led Ihe problem Avas analyzed in detail by Cohn el al (18) In 
additmn cross-reacting substances render the use of antiserum 
unrehab e or a^say of a given antigen. An antiserum dete™Z 
the total of antigenically similar material which is the sum of the 
antigen under consideration and the cross-reacting substances 
Therefore a serum standardized with a nurified V *■ 

would give high values for the concentratioroUhis anlen ^'fbe 
crude inixture containing a cross-reacting antigen 

Furthermore, one should be very careful to make certain he 

using an antiserum analytically that the antigen studied frL 


1 source is the same as that assayed for in another source. For 
example, anti-serum albumin serum has been used to study the 
synthesis of this protein by liver slices without first demonstrat¬ 
ing that the liver protein and serum protein were identical (6). 
It is possible that the liver protein is a cross-reacting substance, 
in which case interpretation of the results would be quite different! 

As a final point, certain differences between molecules are not 
distinguishable immunologically. For example, deamination 
(20) or treatment with formaldehyde under controlled conditions 
(77) results in proteins immunologically indentical with the un¬ 
treated substance. 

under consideration has a known measurable activity, e.g., is an 
enzyme, toxin, virus, or hormone or is characteristically marked, 
i.e., colored, radioactive or with special groupings, the immunologic 
problem becomes greatly simplified. The antigen-antibody reaction 
can be analyzed independently of reactions of antibodies to un¬ 
related substances by use of the 2 markers on the molecules, 
specific activity* and antigenicity. In addition to testing the supei- 
natants for excess antibody or antigen by the presence or absence 
of a precipitate, the supernatants are examined for excess of specific 
activity. If, for example, the specific activity was detected together 
with an excess of precipitating antibody in any given supernatant 
or if an excess of precipitating antigen were present in a supernatant 
with the anti-specific activity, the reaction of an impurity would 
be revealed. 

Let us consider now the question of identification of specific 
activities with given protein molecules. Suppose that we isolate 
an enzyme preparation A which is capable of catalyzing reactions 

Y and Z. On what basis can we decide whether 1 enzyme is cata¬ 
lyzing these 2 reactions or whether 2 separate enzymes are present? 
The immunologic method is one of the most sensitive. Antisera 
are prepared to preparation A. If antibody directed only against 

Y or only against Z is produced in any serum, then Y aiid Z are 
associated with different antigenic species. However, in most 
cases, antibodies will be directed against both. 1 he irnmunologic 
analysis of this system does not involve the precipitates. The 
supernatants are analyzed for anti-F and anti-Z and for Y and 
activities. Two results are possible: either (a) the equn^alence 
points of F and Z are not identical and the separation of F and Z 
is effected immunochemically, or (6) the equivalence pointy lor 
antigen F and antibody Y and antigen Z and antibody Z are 

♦ Any distinctive property of the preparation could be useful, e.g.. absorption due 
to nucleic acid in a nucleoprotcin (22). 



identical and 

remains constant throughout 

the antigen excess zone. 

Considering possibility (a) first, we could say that activities 
Y and Z are associated with antigenically distinguishable species, 
although we could not decide by this test alone whether these 
“distinguishable species” cross-react. Rigorous resolution of 
this problem would require separation of the 2 or the use of reac¬ 
tions in gels, which will be discussed. If possibility (6) were true, we 
could say that Y and Z are associated with identical antigenic 
species. However, this does not mean that the same specific group¬ 
ing is involved in both reactions. One need only recall that tetanus 
and diphtheria antitoxins are indistinguishable not only as anti¬ 
gens but by all physical criteria, yet they are separable by specific 
precipitation with their respective antigens. An analogous situation 
in other protein systems has not been found, but studies of this 
type are not extensive enough. There are interesting examples of 
use of this technique to show that ferritin and the vasodepressor 
substance (42), the a-toxin and lecithinase of clostridia (43), 
the lactase and /3-galactosidase of E. coli (21), and the poly- 
saccharidases acting on S VIII and S III pneumococcus poly¬ 
saccharides (44), respectively, are associated with the same anti¬ 
genic species. There are numerous examples in the literature of 2 
activities supposedly due to a single molecular species; they could 
be tested rigorously by this technique, i.e., ricin and protease (45), 
malic and oxaloacetic decarboxylases (48), maltase and a-phenyl- 
glucosidase (47), oxytoxin and vasodepressor hormones (48), 
and catecholase and cresolase (49). ' 

1 his type of analysis is applicable to the testing of any specific 
property of the molecule. For example, suppose that enzyme 
preparation A contains activity Q, copper and iron and that we 
wish to determine whether this activity is associated with an iron 
or copper protein or neither. Analysis of supernates by the method 
already described would give an equivalence zone for activity Q 
and either copper or iron, or neither at the same noinf. Tf fr... 

This method of analysis depends on the use of a specific com 


billing r6agent, antibody, to resolve 2 different activities into 
distinct physical phases by means of precipitation. The foregoing 
reasoning is applicable in a slightly different manner to systems in 
which this phase separation does not occur. Suppose that we used 
the neutralization reaction, e.g., inhibition of activity without 
precipitation. Two situations may be found. In the first case, 
activity Y is inhibited by a given antiserum but activity Z is not. 
No conclusion can be drawn as to whether the combining groups 
responsible for both activities are identical unless it is known 
that activity Z can be inhibited by an antibody directed against 
it. This latter condition can only be satisfied when both activities 
Y and Z are inhibited by an antiserum preparation. Here, if activity 
Z and activity Y are associated with the same combining groups, 
at the neutralization point of a fixed unit of activity Z, one would 
expect to observe the same residual unit of activity Y with many 
different preparations of antisera. Any other result indicates that 
the 2 activities are associated with different combining groups, 
but not necessarily different protein molecules. This problem can 
become technically very complicated with certain antigens which 
show complex behavior in the neutralization reaction, e.g., phage. 

One of the best examples of the use of these methods is the analysis 
by Cohen and Argobast (50) for the purity of various E. coli B 
phage preparations. They found that antisera to the uninfected 
host E. coli would not precipitate or inactivate the virus activity 
but would remove the non-desoxyribonucleic acid phosphorus 
from the virus preparations. Phage that was further purified after 
treatment with these antisera had all of its phosphorus as desoxy¬ 
ribonucleic acid phosphorus. The specific precipitates contained 
the ribose nucleic acid, and the authors correctly concluded that 
it was not a part of the virus but an unrelated impurity in the 
purified preparations. 

Although we have been using as examples the reactions of 
proteins, the same principles apply to polysaccharides. Kabat and 
his co-workers have developed this type of analysis extensively. 
One example is of great interest. As we have observed, all of the 
antigen is precipitated in presence of excess antibody, so that, 
for a system involving a single antibody, the ratio of the antigen 
in the specific precipitate to the total preparation added gives the 
degree of purity directly. Certain preparations of group A sub¬ 
stance contained 33 per cent glucosamine. Anti-A sera precipi¬ 
tated only about 56 per cent of the glucosamine. The inactive 
material which was not precipitated by anti-A sera was identifie 
0 substance. Of course, impurities not containing glu- 

group U substance. ui cuuioc, , , 7;,T a 

;!mine would not be detected by this method (51). Another 



example is to be found in the work of Perlman and Goebel (52). 

The identification of cross-reacting proteins in a given preparation 
is important in the study of biosynthesis of proteins. On the basis of 
numerous studies (53), we know that the phenomenon of cross¬ 
reaction is due to structurally related substances. The presence in a 
given cell of 2 proteins which cross-react, and are therefore struc¬ 
turally similar, implies an ontogenic relationship between them; 

T, /5-galactosidase (Gz) and structurallv related nrofeins 

extract containing the gSertroportii^of 

at 1,600 Mg N Sd ’ ^ equivalence point G is 

e.g., 1 might be a precursor of the other, or they might have a com 

here describe another ^ ^ gioouiins (55). We will 

tion which can be expected ( 56 , example of a type of reac- 

The /3-galactosidase (Gz) is an adaptive enzyme of the ML strain 


of E. coli and it is induced by a series of substances having the 
unchanged galactosidic; structure. Organisms grown on noninduc¬ 
ing sugars possess only tra(*es of Gz but large amounts of a pro¬ 
tein {Pz) which cross-reacts witli auti-Gssera. Pz is also present in 
adapted cells and, in fact, contaminates the most highly purified 
preparations of Gz. Rabbit antisera to Gz precipitate but do not 
inhibit enzjunatic activity. Pz precipitates about 80-90 per cent 

Fig. 6. —Diagrammatic representation of reactions shown in Figure 5. 
Striateti lower bar represents Gz. Clear upper bar represents Pz. Horizontal 
line represents a fixed quantity of antiserum defined as 100 units (see text for 
explanation). Zone 1, bars 1-3; zone 2, bar 4; zone 3, bars 5-8; zone 4, bar 9; 
zone 5, bars 10-12. 

of the anti-(r 0 , and the antisera absorbed with Pz give the single 
antigen-antibody reaction described previously. 

The interesting property of this system for us is the behavior 
of mixtures of the 2 precipitinogens Gz and Pz. Figure 5 describes 
the quantitative curves of the system, and Figure 6 the diagram¬ 
matic representation. The equivalence point G of the Gz and 
imi\-Gz reactions can be expressed in terms of units of enzymatic 
activity precipitable by a given number of units of antibody. Ihis 
equivalence point is independent of the presence or absence of Pz. 
In other words, Gz has a higher affinity for the anti-(?z than has 


Pz, iiiul in mixtures of the 2, Gz precipitates preferentially. In fact, 
Gz will displace Pz from the soluble complexes in the antigen excess 
zone. The behavior of this system is entirely analogous to the cross¬ 
reaction in horse antisera of S VIII and S III pneumococcus poly¬ 
saccharides (58). As one adds increasing amounts of a mixture 
of Pz and Gz to a fixed, given quantity of antiserum, several re¬ 
gions can be defined (Fig. 6). These zones can be described as 
follows, as we move in the direction toward excess antigen. 

/) All of Pz and Gz, as well as part of the antibody, precipitates. 
There is excess antibody in the supernatant. 

2) All of Pz, Gz and antibody precipitates. This is called equiv¬ 
alence point T and defines the total antigenically similar material 
{Tz). Tz equals the sum of Gz and Pz. 

3) All of the Gz and antibody precipitates, along with part of Pz. 
Excess Pz appears in the supernatant. Gz precipitates preferentially, 
and the unreacted combining groups of the antibody take with 
them part of Pz. Notice that TN in this region is constant. 

4) All of Gz and antibody is precipitated. All of Pz remains in 
the supernatant and is quantitatively separated from Gz. This is 
equivalence point G and defines the quantity of enzymes (Gz) as 
an antigen. 

5) Further additions of antigen result in an excess of Gz in the 
form of soluble complexes, and all of Pz is in solution. 

One can determine the concentration of Pz relative to Gz in 
the following manner. A convenient volume of antiserum is arbi¬ 
trarily defined as equivalent to 100 units of antibody. One hundred 
units of Gz will bring the system to equivalence point G; 100 units 
of total antigen will titrate to equivalence point T. The difference 
between the combining units Tz/wmi vol and combining units Gz/ 
unit vol IS the combining units Pz/umi vol. It should be stressed that 
this titration using 2 specific markers of the molecule is independent 
D antigen-antibody reactions. Depending on the ratio 

Pz/{jz, in remains constant over a variable range, which is wider 
as the ratio IS higher If we did not have the specihc enzymatic 
activity of Gz to guide us, we might conclude on the basis of a 
est of supernatants that we had a single antigen-antibody reaction 

T no antigen or antibody would be 
detectable in the supernatants. Here again, the antigen excess 
sone prov,^ important.. The antigen-antibody complex appeZ 
to be insoluble in excess antigen. The reaction resembrefthlt 
due to a polysaccharide. If neither cross-reacting antigen has a 

S oTtlm "’''y to analysis is the 

e pondMo ZTit'" «"P-rnatant which cor! 

re. ponds to the flat region ot the curve in the antigen excess zone 


is used as a precipitinogen, because a separation of the 2 cross¬ 
reactants might be effected here. The new reaction, in contrast 
to the old one, would not show this flat region, and the new pre¬ 
cipitinogen should behave like a cross-reacting substance, pre¬ 
cipitating less antibody/unit antigen than the homologous antigen. 
On the other hand, if the 2 cross-reacting proteins had identical 
affinity, it would be impossible, in the absence of another specific 
marker, to distinguish them by these immunochemical methods 
alone. The use of reactions in gels might resolve them. 

The behavior of Pz and Gz suggests that a similar system of 
cross-reacting proteins is present in preparations which show the 

same type of quantitative precipitin curves. Here we can only 
speculate. Jager et ol. (59) and Ivabat et ol. (60) have suggested 
that 72 -globulin components might cross-react with other serum 
proteins. Cohn, Deutsch and Wetter (18) have suggested that their 
results do not rule out the possibility of the reactions of unrelated 
components common to various 7 -globulin preparations. However, 
the form of several of the curves published by the latter worl^rs 
supports the suggestions of Jager, Rabat and co-workers. For 
example, the reaction of yn.-globulin with anti- 7 --globulin sera 
could be due to the reaction of 2 cross-reacting proteins, for t e 
curves show that the precipitinogen is remarkably insoluble m 
excess of antigen, quite unlike ordinary protein-antiprotein reac- 



tions. This was not explained. Similar findings have been reported 
for reactions between chicken whole egg white and anti-conalbumin 
sera (17) and milk whey proteins and anti-iS-lactoglobulin sera (23). 
These interesting systems deserve further investigation. 

Flocculation reaction .—This antigen-antibody reaction, first 
interpreted in quantitative terms by Pappenheimer and Robinson 
(16) and represented by the curve in Figure 7 is relatively simple 
to analyze. It is characterized by precipitation in a narrow region 
and a broad equivalence zone. Curve A represents the reaction 
between a pure antigen and its flocculating antibody. Since the 
total antibody is a constant and all antigen and antibody are pre¬ 
cipitated in the equivalence zone, a straight line represented by 
equation (4) is obtained for this region. 

TN = PX + A (4) 

where X and A are defined as in equations (1), (2) and (3) and P 
is the ratio of specifically precipitable antigen N to total N added, 
in other words, the percentage purity in terms of nitrogen. 

Curve A of Figure 7 has been dra^vn to permit comparison 
with curve A of Figure 3. Total antibody is chosen as 1.0 mg N 
and P = 4 at the point of optimal flocculation. A 100 per cent 
pure protein was used as precipitinogen and the slope of the line 
is therefore 1.0. Curves B and C show what happens when pre¬ 
cipitinogens, 50 and 25 per cent pure in terms of N, are treated with 
this antibody; the slopes of the lines in the equivalence zone are 
0.50 and 0.25, respectively. All of these lines extrapolate to 1.0 mg 
N, which is the total antibody. 

To assay for a given antigen in a mixture, it is enough to deter¬ 
mine TN at any point in the equivalence zone. Since X and A 
are known, equation (4) (‘an be solved for P, which is the percentage 
purity. This technique has been used to determine the purity 
of diphtheria toxoids (14). 

A curve of the type shown in Figure 7 provides clear evidence 
that a single antigen-antibody reaction is involved. This means 
simply that antibody reacting in the system precipitates a single 
antigenic species. The fl()cculating system cannot be used to detect 
small amounts of impurities, as the determination of the slope of 
the line IS only good to about 2 per cent, so that given protein 
prepar^ions between 96 and 100 per cent pure will be indistinguish¬ 
able, t However, the flocculating system, in the equivalence region 
separates the precipitinogen under investigation from its impuri- 

in the supernatants. Rabbit antisera prepared 

flocculating antibody is directed against a 


against this same antigen might contain antibodies against im¬ 
purities, which could he detected and in some cases even assayed 
for in the supernatants in the equivalence zone of the flocculation 
reaction. Phis was aid.iially done to detect impurities in certain 
conalbumin preparations (17) which gave a curve of the floccula¬ 
tion reaction with a slope of 1 in the equivalence zone. The latter 
technique is extraordinarily sensitive, and one can detect as little 
as 0.1 per cent impurity with no difficulty. 

Another extremely useful property of the flocculation reaction 
is that precipitation takes place in zones. This makes it possible (1) 
to obseiwe the reactions of minor components which may pre¬ 
cipitate in other zones, and (2) to separate these reactions with a 
minimum of difficulty. 

An example is given in Figure 7. Our hypothetical flocculating 
antisemm showed 2 zones of optimal flocculation with a given 
antigen preparation. The curves obtained on quantitative analysis 
ran into each other, as shown by curves B and D. The straight 
line portion of curve B was sufficiently distinguishable to show a 
slope of 0.50 and that of the 2d curve D was 0.067. This means 
that precipitinogen B is present in 50 per cent and D in 6.7 per cent 
of the total preparation, and extrapolation of these lines to zero 
shows that 1.0 mg antibody N is directed against precipitinogen 
B and 0.50 mg against D. A rather crude separation such that 
precipitinogen B is purified to 60 per cent and D reduced to 
3 per cent would completely separate these systems immunologi- 
cally. It is obvious that even if the 2 curves showed 90 per cent 
overlap, the complex reaction could be observed and simple tech¬ 
niques that would change the relative concentrations of the 
precipitinogens by as much as 10-20 per cent might separate the 
curves sufficiently for adequate analysis. Experimentally, the best 
examples of this type behavior are the reactions of chicken oval¬ 
bumin and conalbumin with horse antisera (17) and the double 
zones of flocculation found with antitoxin to scarlet fever toxin (61). 
It is obvious that one can determine the concentration of a protein 
which has not yet been isolated if a flocculating antibody to it is 
available. This was done successfully for diphtheria (16), tetanus 

(62) and scarlet fever toxins (63). 

Despite the great usefulness of this system,! it is rarely used, for 

t One other niece of information that may sometimes be obtained from analysis 
of the snSific Secfpitate at the point of optimal flocculation s worth mentioning. 

If III rSmesrahe molecula? formula of the rt^^Xt^T^e anUgeT^^^^^ 
of antibody to 1 molecule of antigen, the molecular weight 

calculated. The molecular weight of horse pseudoglobulin J^is calcu a 

tinn was successfully applied to diphtheria toxin (73) and ovalbumin (65). 

- tIoVXooo. At 



several reasons, (1) The system has only been observed in sera 
from horses, "whose care makes laboratory use difficult. (2) Often 
tlie period of immunization must be long, and large quantities of 
antigen are retiuired. ( 3 ) 11 ' is not always possible t() obtain floccu¬ 
lating antisera which giA'e clear-cut results. 

It is unfortunate, nevertheless, that more work has not been done 
on elucidating the factors which differentiate precipitin from the 
flocculation-type antibody. Convenient production of the latter 
type would be of enormous value to immunology. Flocculating anti¬ 
bodies have been described for a large series of proteins, diphtheria 
(16), tetanus (62), scarlet fever (63) and botulinus toxins (64, 30), 
chicken ovalbumin (65) and conalbumin (17), human and rabbit 
serum albumins ( 66 , 67) and hemocyanin ( 68 ). 

To obtain the typical curves of the flocculation reaction, high 
titer antisera must be used. Even in the reaction of diphtheria 
toxin, which has received most extensive study, weak antisera 
give apparent precipitin-type curves. Other aberrations are found 
when nonspecific precipitation (69) or combined reactions of floccu¬ 
lation and precipitin types occur (70, 71). In addition, there is 
preliminary evidence that certain protein antigens, e.g., 7 -globulin, 
induce precipitin antibody in the horse (72). One cannot expect 
that every horse immunized with a protein will give antiserum 
which will show a typical flocculation reaction. 

In addition, it should be mentioned that horse antisera to poly¬ 
saccharide give typical precipitin-type reactions which might 
be due to the peculiar characteristics of the polysaccharides 
(large linear molecules with repeated combining groups) or to the 
special properties of polysaccharide antibodies which are of much 
higher molecular weight and greater insolubility than the floccu¬ 
lating antibodies. With respect to this question, the interesting 
observation of Grabar and Staub (74) should be mentioned. 
Horse antisera against the glycoprotein of the anthrax bacillus 
were shown to give a precipitin-type reaction with this precipitino- 
pn. However, the pseudoglobulin fraction reacted to give a floccu- 
lation reactip. It seems as though this polysaccharide-protein 
induced pe formation of both types of antibody which were sena- 
rable with euglobulm and pseudoglobul in fractions. 

appenheimer introduced the interesting hypothesis that the 
floeeulatmg antibody differed from the preefp'tin antibX I 

to-mined ffom'LdfmSati»n*1ltuSon'dta"w'f^^ H f'”'*■''>'■*‘‘ 01 ' »» de- 
logic value appeared low Two years The immuno- 

tallizecl conalbumin. and froTn fts Tron conSn Earner and Weber (78) cryL 

the molecular weight to be 70,200-76 600 ^rvatue capacity reported 

immunologic one. o.ouu, a value in striking agreement with the 


that the former had both combining groups on the same side of the 
molecule and that 1 site interfered with the reaction of the other 
(73). Kendall has postulated the presence of 2 combining groups 
with different reactivity and derived a mathematical relationship 
which describes the flocculation reaction remarkably well (75). 
There are other possible explanations which can be tested. 

The flocculating system might be due to a mixture of precipitat- - 
ing and nonprecipitating antibodies. The latter inhibit precipita¬ 
tion in large antibody excess and in the equivalence zone behave 
much as do the univalent antibodies in the cross-reaction between 
pneumococcus polysaccharide type III and anti-type VIII sera 
(76); i.e., the linear portion of the curve of a flocculation would re¬ 
sult from combined complete precipitation of precipitating and 
nonprecipitating antibodies. Evidence in support of this theory is 
that horse antisera have large amounts of nonprecipitating anti¬ 
body in the early stages of immunization, which means that the 
animal does synthesize this type of antibody. This hypothesis pre¬ 
dicts that correct artificial mixtures of precipitating and non¬ 
precipitating antibodies would give a typical flocculation reaction 
with the corresponding precipitinogen. 


The reviewers have felt that the author oversimplified the discussion of 
the flocculation reaction, for recently 2 papers (69, 79) have appeared in 
which certain quantitative aspects of the flocculation reaction as described 
here have not been confirmed. The reviewers have considered these papers 
of sufficient importance to merit an extended discussion and have sug¬ 
gested that they be analyzed in detail. 

All workers agree that the flocculation reaction is characterized by solu¬ 
bility of the antigen-antibody complex in both antigen and antibody ex¬ 
cess; therefore the reaction takes place in a zone. In fact, many horse anti¬ 
tetanus and scarlet fever antitoxins react in more than 1 zone, and the 
problem is to find which zone is due to the toxin and which to unrelated 
material, a question only resolved by analyzing the supernatants for dis¬ 
appearance of specific activity (61). Precipitation in a zone has also been 
confirmed by investigations on the reaction m gels (see Oudin, this vol- 

The discussion which follows involves the “straight line” portion of 
curve from which the percentage purity of a given antigen can be deter¬ 
mined. Since this question has been raised most pointedly by Pope et at. 
(79) with regard to the diphtheria toxin-antitoxin reaction, this system and 
their recent report on it will be analyzed. These workers find that the reac- 
Uon takes plaL in a zone but state that they f 

single curve of the type reported by Pappenheimer and Robinson (1937) 
or Cohn and Pappenheimer (1949) for horse antitoxin. This refers to the 
“straight line portion” of the curve. 


It has been known since the time of Ehrlich and confirmed by many 
workers that the reaction between diphtheria toxin and horse antitoxin 
exhibits an exceptional behavior, in that the difference between the L+ 
dose and the Lo* dose of toxin is not equal to 1 minimum lethal dose but 
is much greater (about 30 times) and that in general the L+ and Lo doses 
are greater than the Lf dose determined at the point of optimal floccula¬ 
tion. The uninitiated reader need not fatigue himself over these units, for 
the aigument should clarify itself without them. In terms of quantitative 
immunochemistry the discrepancy between L+ and Lo, taken together 
with the large Danysz effect observed after all of the antitoxin is com¬ 
bined, signifies that the reaction between diphtheria toxin and horse antitoxin 
is characterized by a broad equivalence zone, t 

With some hindsight, then, one can state that it should not liave been a 
surprise that Pappenheimer and Robinson found a linear relationship be¬ 
tween toxin N added and total N precipitated, since it was implicit in the 
already known finding that with horse antitoxin there is a large series of 
antigen-antibody mixtures over which all toxin and antitoxin is precipi¬ 
tated. In fact, had Pappenheimer and Robinson uncovered any other rela¬ 
tionship the explanation of the relationship between the various units 
(L-h, Lo, Lf) of toxins and antitoxins would have remained obscure. 
In 1937, Pappenheimer and Robinson reported that this finding was re¬ 
producible in 6 different sera, and since then many laboratories have con¬ 
firmed their observations. Furthermore, similar results have been obtained 
with at least 6 other antigens, as mentioned in the text. 

It might be pertinent to point out that it would not have been necessary 
to repeat their curve with many sera. It suffices to have one system which 
follows the “ideal” flocculation reaction to draw precise conclusions con¬ 
cerning the percentage purity of various toxin preparations. The fact that 
one can find systems, as have Pope et al., which for one reason or another 
do not show the “ideal” reaction is of little significance from the point of 
view of determining homogeneity of proteins. The recent reports of Van 
Ileyningen et al. (69) and Pope et al. (79) do not deny the essential charac¬ 
ter of the flocculation reaction. They demonstrate, rather, that it can be 
complicated by unrelated factors. It is obvious, therefore, that if the 
straight line region were present, it could easily be masked by nonspecific 
precipitation (69) or by overlapping precipitation of unrelated systems 
(79). But the danger here is no greater nor is it subject to different controls 
tian in the precipitin system; with due regard to similar complications, 
rabbit antisera have been used to assay many systems for given antigens. 
Since Pope et al (79) do not report on analyses of the supernatants for the 
reactions the y studied, it is impossible to find out over what region of their 

"Tii'i,"'!'"’ ff to™ o oSrved (|o)"'“ “0 g 

withilf tSira',d’‘l"i?et‘?d 

weight, will. 0,1 the average, khl that guine^^^^^^ =50 g 

lieen p“ntero«tha1'rn?tsr,"e"“„t1r„S^^^^^^^ ‘"<1 it 

of the aocculation reaetioo and would not apply' to thS pJecS rTuon ““ 


curves the straight line portion (equivalence zone) could be expected and 
thereby calculate how much is contributed by unrelated systems. 

Provided that one has a single antigen-antibody flocculation reaction 
with a broad equivalence zone and that non-specific precipitation is not 
occurring, the calculation of the percentage of given antigen in the prepa¬ 
ration under analysis is an obligatory deduction. One of the arguments 
against this calculation, as advanced by Pope et al, is that the purity of 
their best preparations of toxin or toxoid, in flocculation units (Lf/mg 
N) is greater than the value of 2,170 predicted by Pappenheimer and Rob¬ 
inson (16) from their analysis of the flocculation reaction. The outsider to 
this discussion might think that the flocculation unit (Lf) is akin to a phys¬ 
ical constant such as the value for the speed of light. The unit depends, 
rather, on comparison of given horse antitoxin, designated by international 
agreement as a standard with another horse antitoxin intended for the 
routine analytical work. The in vitro comparison (optimal flocculation 
against a given toxin) is subject to significant errors of technique and to 
the danger of choosing a serum in which the flocculation of unrelated sys¬ 
tems masks the reaction of the toxin, as is discussed by Pope. The in vivo 
comparison has been well analyzed by Jerne (81), whose conclusion is 
quoted: “Standardization of antitoxin is shown to have been based on 
assumptions. The authority bestowed upon the International unit for 
diphtheria antitoxin has retarded the development of science in this field.” 
This author points out that 2 antitoxins could have the same unitage as 
determined by flocculation reaction and differ markedly when compared 
by skin neutralization test (in vivo), and vice versa. 

To convince oneself as to how variable is the unitage, it is necessary only 
to send the same toxin preparation to a number of laboratories and ask 
them to determine the Lf/mg N of the preparation. In our experience, 
variations of the order of magnitude involved here have been obtained. 
In fact, the most valid method available, as pointed out by Pappenheimer, 
is that of selecting antitoxins which show the “ideal” flocculation reaction 
and employing them to standardize toxins in terms of specifically precipi- 
table nitrogen. It is highly probable that the difference between the pre¬ 
dicted value (Lf/mg N) of purified diphtheria toxin and the higher values 
found by other workers may be a question only of how the unitage of the 
various antitoxins employed was determined. The real question here is 
not how many Lf/mg N are possessed by a purified toxin, but rather what 
percentage of a given preparation is specifically precipitable as toxin. If 
the problem is to be decided of how many Lf/mg N the purified toxin pos¬ 
sesses, all workers must use the same antitoxin and agree on its unitage. 
It is obvious that if the predicted value were lower than the value deter¬ 
mined for purified toxin and this discrepancy is not due to errors of unitage, 
one would be forceil to conclude that toxin was synthesized during the 
course of its fractionation or that in tlie flocculation reaction more toxin is 
specifically precipitated than is added. Pope’s own data support the vie\v 
tLt these differences are simply due to the designatimi 

2 antitoxins used, i.e., Pappenliei.nei ’s in 1937 and Pope s in 95 - W‘I' 

3 different pepsin-treated antitoxins and Pappenheimer s purified toxin, 



Poi)e et al. determined the N content/toxin-antitoxin unit by analyzing 
the specific precipitate formed at the point of optimal flocculation. They 
found 22.3, 23.7 and 21.3 X IQ-* mg N, respectively, with these 3 antitoxic 
preparations and point out that the value is quite different from the value 
of 16.1 X 10“'* mg N reported by Pappenheimer and Robinson. The ex¬ 
cellent agreement between the values found with the first 3 antitoxins by 
Pope et al. would be unlikely if appreciable amounts of unrelated specific 
precipitation or nonspecific precipitation had taken place at the point Ox 
optimal flocculation, for the chances of having 3 horses respond identically 
to all antigenic components is small. Therefore, with these 3 pepsin-treated 
antitoxins. Pope et al. were probably studying a single antigen-antibody 
reaction. For this reason, the likely explanation of the different results, 
Pappenheimer’s and Pope’s, is that of differences in unitage.J 

Furthermore, the sample of Pappenheimer’s purified toxin analyzed by 
Pope, had been sent to him in lyophilized form in 1937. Since the activity 
of lyophilized toxin is known to fall off unless dissolved quickly after open¬ 
ing, it is difficult to know what changes could have occurred in the prepa¬ 
ration. In carrying out the analysis of TN/unit at the point of optimal 
flocculation. Pope, however, used the value 2,170 Lf/mg N determined in 
1937 by Pappenheimer for this preparation. 

In a personal communication, Pappenheimer states that he has retitrated 
recently 1 of the antitoxins extensively studied in 1939 against a new stand¬ 
ard flocculating antitoxin and found that the old serum had apparently 
gained 25 per cent in unitage. In the light of these findings, it would be of 
great interest to know the value (Lf/mg N) determined by Pope for the 
purified toxin of Pappenheimer. 

Changes in purified toxin on reconstitution following storage in the 
dried state, therefore, in addition to uncertainties in the unitage of differ¬ 
ent antitoxins, might in themselves account for the discrepancies between 
the value (Lf/mg N) assigned to purified toxin by Pappenheimer and Rob¬ 
inson and the results reported by Pope et al. 

Continuing their argument. Pope et al. claim that with certain of their 
antitoxins, Pappenheimer’s purified toxin showed 14 components when 
tested by the technique of precipitation after diffusion in a gel. This toxin 
preparation had been analyzed to be 93.5 per cent pure by use of the floc¬ 
culation reaction (14); i.e., 93.5 per cent of the preparation was specifically 
precipitable by antitoxin. If, as Pope has done, the precipitation in gels is 

most of the antibody is directed 
against the atoxic proteins, it is quite possible that one would see the 13 

corresponding to the reactions of the same number ofanti- 
gensmUh^6.5 per cent of impurities. The number of boundaries is irrele- 

othe^dfffer'lnLt of units. An- 

aiid extent of peptic treatment of the antitox'inr^From^tl^^^^ method 

eus.sion it would have been more nertinenf 1 . P ^ of view of this dis- 

toxin.s. Pappenheimer and Robinson showed tlmt wU*!? 4 untreated anti¬ 
toxins and .3 different toxin preparations Tlfev-xlm • , -1 untreated anti- 

antitoxin. The 1 different resnVt xa.tuinJ. i ^ obtained identical values of TN /unit 

•Mitoxic globulin, 

preparation with 3 others obtained 14 years ater 11 commercial 

o preparation and extent of digestion are unstated. ^ ^ techniques 


vant to the question of the percentage of toxin in the preparation. For ex- 
arnple, as an extreme case, if one had antibody only against the atoxic pro¬ 
teins, and a toxic preparation of 99 per cent purity was tested against it 
in a system in gels, one might see 13 antigen-antibody reactions, none of 
which is due to the major component, the toxin, and conclude, therefore, 
that the preparation was grossly impure. Bowen (82) recently reinvesti¬ 
gated the reaction between toxin and horse antitoxin using the technique 
of diffusion in gels. He confirmed Pope’s observations that many sera do 
possess detectable antibody against atoxic proteins. However, he carried 
the observation further by attempting a semiquantitative analysis of the 

system. Bowen demonstrated clearly that by using very high concentra¬ 
tions of toxin (1,000-3,000 Lf/ml), as Pope has done, one can demonstrate 
.a large number of reactions due to atoxic protein impurities in the purified 
preparations, but that these impurities may amount to a very small per¬ 
centage of the total antigenic material present. 

The findings reported by Pope et al. on precipitation in gels have not 
been uniformly encountered. For example, Oudin, in the chapter which 
follows, discusses the reaction in gels between diphtheria toxoid and horse 
antitoxin prepared at the Pasteur Institute and reports the finding of a 
single boundary. With certain antitoxins Ouchterlony (83) found 3 bound¬ 
aries, 1 distinctly due to the toxin itself and the other 2, very faint, due to 
unrelated systems. The author, in an unpublished investigation of atoxic 
diphtheria proteins, tested about 30 high titer horse antitoxins pre¬ 
pared commercially in the United States but did not find one with suf¬ 
ficient antibody against the atoxic proteins to be useful in following the 
isolation of these antigens. Pope et al. apparently have worked with sera 
which contain larger amounts of antibody against the atoxic proteins. 
These complex systems, however interesting, cannot be used to analyze 
(1) the properties of the flocculation reaction or (2) the purity of various 
toxin preparations. Simpler systems must be looked for and employed. 

Pope et al. remark that their observations show that diphtheria toxin 
has not been obtained as a pure substance. If multiple, rigid criteria are 
applied, the same can be said for most highly purified proteins. In any 
case the present author is of the opinion that the preparations studied by 
Pappenheimer were sufficiently pure to enable a valid determination of 
molecular weight and chemical composition to be made, and that these de¬ 
terminations are not invalidated by more recent observations. In addition. 
Pope et al. were apparently unaware of the work of Agner (84), who con¬ 
firmed Pappenheimer’s observations on the purification of toxin. Agner 
isolated a toxin preparation at least 99 per cent specifically f> 

which moved as a single boundary on electrophoresis and which had toxic 
activity of 55 MLD/Lf and 2,200 Lf/mg N. Pope et al. have used 
immunologic criteria only and have not introduced measurement of tox¬ 
icity in their analyses. Finally, it is appropriate to point 
preparations have been obtained which are pure by all f 
criteria (solubility, electrophoresis, etc.) and winch contain 2,10^2,200 
bf/mu N comparable to the equivalent toxin preparations (85). bince 
toxin and toxoid are immunologically indistinguishable, there is no reason 


to exfject that isolated samples of toxin will possess values of Lf/mg N 
markedly difYerent from the corresponding toxoid. 


1. Heidelherger, M., and Mayer, IM.: (Quantitative chemical studies on 

complement: IV. Addition of human comi)lement to specific precipi¬ 
tates, J. Exper. Med. 75: 285, 1942. 

2. Danysz, J.: Contribution i I’dtude des proprietes et de la nature des 

melanges des toxines avec leurs antitoxines, Ann. Inst. Pasteur 16: 331, 

3. Scholander, P. F.: Microburette, Science 95: 177, 1942. Scholander, 

P. F.: Edwards, G. A., and Irving, L.: Improved microburette, J. Biol. 
Chem. 148: 495, 1943. 

4. Bonner, W. (4., and Branting, B. F.: The composition of hydrochloric 

acid of constant boiling point, J. Am. Chem. Soc. 48: 3093, 1926. 

5. Heidelherger, M., and MacPherson, C. F. C.: (Quantitative microestima¬ 

tion of antibodies in the sera of man and other animals. Science, 97: 
405; 98: 63, 1943. 

6. Peters, T., Jr., and ,\nfinsen, C. B.: Net production of serumalbumin 

by liver slices, J. Biol. Chem. 186: 805, 1950. 

7. Barbour, II. G., and Hamilton, W. F.: The falling drop method for deter¬ 

mining specific gravity, J. Biol. Chem. 69: (525, 1926. Lowry, (4. H., and 
Hunter, T. H.: The determination of serum protein concentration 
with a gi'adient tube, J. Biol. Chem. 159: 465, 1945. Sunderstrom-Song, 
K., and Song, H., Jr.: Enzymic histiochemistry: XXIX. Dilatometric 
mieroestimation of peptidase activity, Compt. rend. Trav. Lab. Carls- 
berg, sdr. chem. 21: 315, 19.38: Mikrochem. acta 3: 210, 1938. 

Kunkel, H. G., and Ward, S. M.: The immunological determination of 
human albumin in biological fluids, J. Biol. Chem. 182: 597, 1950. 
Lisen, H. N.. Liltraviolet absorption spectroscopy of immune pre¬ 
cipitates, J. Immunol. 60: 77-83, 1948. Gitlin, 14.: the use of ultraviolet 
absorption sj)ectroscopy in the quantitative precipitin reaction J 
Immunol. 62: 437-451, 1949. 

Libby, R. L.: A photoreflectometer, an instrument for the measure¬ 
ment of turbid systems: A new ami rapid quantitative technic for the 
determination of the iiotency of types I and II antipneumococcal 
serum, J. Immunol. 34: 71 and 269, 1938. 

Rabat, E ^ ^ ^^^P^rimenlal Immunorhemistry 

(Sinmgheld, Ill.: C harles C Thomas, Publisher, 1948). 

.oweleur J (chI.): Techniques de labomtoire Masson & Cie, 1947 ). 
(Gralnr)*’ ^ unmunocliimitiue quantitative et son einploi 

"fvf’Tn"’ ''""‘‘“'I’ '"• reaction l,etween 

tj ie HI pneumococcus Dolysaecharide and homologous antibody 

Espei. MeTuhSh'mC''''' J- 


8 . 


10 . 

11 . 

12 . 





dlpluhc:irtv)‘'’'Nn'"™''- ‘irdtotitative study of the 

ul!.,.. ,! . .4'■‘■actum in the sera of various species 

including man, J. Immunol. 63: 291 1949 
Heidelherger, M., and Kendall, F. The 
line egg-albumin and its homologc 

reaction between crysfal- 
?ous antibody, J. Exper. Med. (52: 

l>appcnhei,ner, A. M., Jr., an,I llobiuson. E. S.: A quantitative sltuiy 





20 . 

21 . 

22 . 












34 . 


of the Ramon diphtheria flocculation reaction, J. Immunol 32- 291 

Cohn, IM., etter, L. It., and Deutsch, H. F.: Immunological studies 
on egg-white proteins: I. Precipitation of chicken ovalbumin and 
conalhumin by rabbit- and horse-antisera, J. Immunol. 61: 283, 1949. 

Cohn, IVI.; D(»utsch, H. I., and Wetter, L. It.; Biophj'sical studies of 
blood ])lasma proteins: XIII. Analysis of immunological heterogeneity 
of human gamma globulin fractions, J. Immunol. 64:381, 1950. 

.\dams, M. II.: The reaction between the enzyme tju-osinase and its 
specific antibody, J. Exper. Med. 76: 175, 1942. 

Maurer, P. H., and Ileidelberger, M.: The deamination of crj'stalline egg 
albumin; III. Quantitative immunological studies on crj'stalline egg 
albumin and its denatured and deaminated derivatives, J. Am. Chem. 
Soc. 73: 2076, 1951. 

Cohn, M., and Monod, J.: Purification et propriet4s de la beta-galacto- 
sidase (lactase) Escherichia coli, Biochem. et biophys. acta 7: 153, 

Staub, A. M.; Prudhomme, It. O., and Grabar, P,; Recherches immuno- 
chimiques sur la bacterie charbonneuse: V. Les nucleoprot4ides, 
Ann. Inst. Pasteur 72: 58, 1946. 

Deutsch, H. F.: Immunochemical studies of beta-lactoglobulin, J. Biol. 
Chem. 185: 377, 1950. 

Wetter, L. R., and Deutsch, H. F.: Immunological studies on egg-white 
proteins: III. Quantitative immunological studies of ovomucoid. Arch. 
Biochem. 29; 399, 1950. 

W'etter, L. R., and Deutsch, H. F.: Immunological studies on egg white 
proteins; IV. Immunological and physical studies of lysozyme, J. Biol. 
Chem. 192: 237-242, 1951. 

Heidelberger, M., and Kendall, F. E.: A quantitative theory of the pre¬ 
cipitin reaction: IV. The reaction of pneumococcus specific polysac¬ 
charides with homologous rabbit antisera, J. Exper. Med. 65: 647, 1937. 

Rabat, E. A.; Baer, H.; Day, R. L., and Knaub, V.: Immunochemical 
studies on blood groups: XL Species differences among blood group A 
substances, J. Flxper. Med. 91: 433, 1950. 

Heidelberger, M., and Kendall, F. E.: The role of multiide reactive 
groups in antigen-antibody union as illustrated by an instance of 
cross precipitation, J. E.xper. Med. 59: 519, 1934. 

Malkiel, S., and Stanley, W'. M.; Immunochemical studies on tobacco 
mosaic virus: I. The relation with homologous rabbit antiserum, J. 
Immunol. 57: 31-42, 1947. 

Lamanna, C., and Doak, B. W’.: The antigen-antibody reaction with 
type A botulinal toxin, J. Immunol. 59: 231-247, 1948. 

MacPherson, C. F. C., and Heidelberger, M.: Denatured egg albumin: 
III. Quantitative immunochemical studies on crystalline egg albumin 
denatured in various ways, J. .\m. C’hem. Soc. 67:585, 19-15. 

Ileidelberger, M., and Rabat, E. A.: The dissociation of antibody from 
pneumococcus specific precipitates and specifically agglutinated 
nneumococci, J. Exper. Med. 67: 181, 1938. 

llaurowiU, F., .aii.l Schwerin, P.: Tlie specihcty of antibodies to antiKens 
containing two different determinant groups, J. Immunol. 4i. 111-11.1, 

1Q43 . . 

Dean II R.; Taylor, G. L., and Adair, M. E.: The precipitation reac¬ 
tion: Experiments with an aiiti.serum containing two antibodies, 
J. Hyg. 35; 69,1935. 
























Kendall, F. E.: Studies on serum proteins: I. The identification of a 
single serum globulin by immunological means, J. Clin. Invest. 16: 921, 

Culbertson, J. T.: The role of the precipitating antibody in the removal 
of intravenously injected antigen, J. Immunol. 28: 279, 1935. 

Weil, R.: Characteristics of the jirecipitin reaction, Proc. Soc. Exper. 
Biol. & Med. 13:200, 1916. 

lleidelberger, M., and Kendall, F. E.: Conditions for the quantitative 
precipitation of antibody in horse sera, J. Exper. Med. 61: 559, 1935. 

Heidelberger, M.; MacLeod, C. M.; Markowitz, H., and Roe, A. S.: 
Improved methods for the preparation of the specific polysaccharides 
of pneumococcus, J. Exper. Med. 91: 341-349, 1950. 

Scherp, II. W.: Studies on meningococcal infection: XI. A quantitative 
study of the precipitative reaction between type I polysaccharide 
and anti-meningococcal horse sera, J. Immunol. 37: 469, 1939. 

Engel, L. C., and Randall, R.: The quantitative estimation of egg and 
chicken proteins in equine encephalomyelitis vaccines, J. Immunol. 55: 
325-328, 1947. 

Mazur, A., and Shorr, E.: The identification of hepatic vasodepressor 
substance with ferritin, J. Biol. Chem. 176: 771, 1948. 

MacF''arlane, M. G., and Knight, B. C. J. G.: The biochemistry of bacte¬ 
rial toxins: I. The lecithinase activity of Clostridium welchii toxins, 
Biochem. J. 35: 884, 1941. 

Shaw, M., and Sickles, G. M.: Production of specific pneumococcus 
carbohj'drate-splitting enzymes in media to which the specific sub¬ 
strate was not added, J. Immunol. 64: 27, 1950. 

Le Breton, E., and Moule, Y.: Sur le mecanisme des toxalbumines: Gas de 
la ricine. Bull. Soc. chim. biol. 31: 94, 1949; compt. rend. 229: 305, 

Korkes, S.; Campillo, A., and Ochoa, S.: Isolation and properties of an 
adaptive “malic” enzyme from Lactobacillus arabinosus, J. Biol. Chem. 
187:891, 1950. 

Hestrin, S.: Carbohydrates in Saccharomyces haploid stocks of defined 
genotype: I. Fermentation and hydrolj'sis of alpha-glucosides by yeast 
6233, Arch. Biochem. 29: 315, 1950. 

Chow, B. F.: Purification and properties ol certain protein hormones 
Adv. Protein Chem. 1: 153, 1944. ’ 

Mallette, h. M., and Dawson, C.: On the nature of highly purified mush- 
^ room tyrosinase preparations. Arch. Biochem. 23:29, 1949. 

Cohen, &., and Argobast, R.: Chemical studies on host virus interactions: 
\ I. Immunochemical studies on the purity of concentrates of various 
bacterial viruses prepared by differential centrifugation procedures 
J. Exper. Med. 91: 607, 1950. 

Bendich, A.; Kabat, E. A., and Bezer, A. E.: Immunochemical studies 
on blood groups: III. Properties of purified blood group A substances 
Irom individual hog stomach linings, J. Exper. Med. 83, 485, 1946. 
erlman, E-, and Goebel, W. F.: Studies on the Flexner group of dysen- 

Kabat, E A.; Heidelberger, M., and Bezer, A. E.: Study of the ourifica 
tiori and properties of ricin, J. Biol. Chem. 167: 629, 1947 
reffers, II. 1Moore, D. H., and Heidelberger, M.: Antigenic properties 






() 0 . 

01 . 

02 . 




00 . 










of liorse serum fractions isolated by electrophoresis and ultracentrifu¬ 
gation, J. E.xper. Med. 75: 135, 1942. 

Cohn, M., and Torriani, A. M.: Etude immunochimique de la biosyn- 
these adaptative d’un enzyme: La lieta-galactosidase (lactase)'d’- 
^ Eschenchia coli, Compt. rend. 232: 115, 1951. 

C ohn, AI., and Torriani, A. AI.: Alanuscilpt in jireparation. 

Ala^ei, AI., and Heidellxa'ger, AI.: \Tlocity ol (toml)ination of anti¬ 
body with specific polysaccharides of pneumococcus, J. Biol. Cdiem 143- 
507, 1942. 

Jagei, B. \ ., Smith, Ij. L.; Nickerson, AI., and Brown, D. AI.: Immuno¬ 
logical and electrojihoretic studies on human gamma globulins, J. 
Biol. Chem. 170: 1177, 1948. 

Rabat, E. A., and Alurray, J. P.: A comparison of human gamma globu¬ 
lins in their reactivity with rabbit anti gamma globulin by the quanti¬ 
tative precipitin method, J. Biol. Chem. 182: 251, 1950. 

Rao, S. S., and Aloloney, J. P.: Estimation of potency of scarlatinal anti¬ 
toxin by combined flocculation and rabbit skin test method, J. Im¬ 
munol. 04:57, 1950. 

AIalone.y, P. S., and Hennessy, J. N.: Purification of tetanus toxoid, 
Biochem. J. 30: 544, l‘)42. 

Ilottle, G. A., and PapiKUiheimer, A. AI., Jr.: Quantitative study of 
scarlet fever toxin-antitoxin flocculation reaction, J. Exper. Aled. 74: 
545, 1941. 

Abrams, A.; Kegeles, G., and Hottle, G. A.: The purification of toxin 
from Clostridium hotulinum type A, J. Biol. Chem. 104: 03, 1940. 

Pappenheimer, A. AI., Jr.: Anti egg albumin antibody in the horse, J. 
Exper. Aled. 71: 203, 1940. 

Gitlin, D.; Davidson, C. S., and Wetterlow, L. H.: The quantitative 
estimation of serum albumin in human body fluids by direct titration 
with specific horse antiserum, J. Immunol. 03, 415, 1949. Gitlin, D., 
and Edelhoch, H.: A study of the reaction between human serum alt)u- 
min and its homologous equine antibody through the medium of 
light scattering, J. Immunol. 00: 07-77, 1951. 

Ileidelberger, AI.; Treffers, II. P., and Freund, J.: Antibodies to rabbit 
serum albumin and their reaction with antigen, J. lixper. Aled. 80: 

83, 1947. 

Hooker, S. B., and Boyd, W. C.: Equine anti-hemocyanin, Ann. New 
York Acad. Sc. 43: 107, 1942. 

van Ileyningen, M . E., and Bidwell, E.: The biochemistry of the gas 
gangrene toxins: IV. The reaction between the alpha-toxin (lecithinase) 
of Clostridium loelchii uml its antitoxin, Biochem. J. 42: 130, 1948. 

Perez, J. .1.: Etude des scrums de chevaux anti-inotcides: Influence 
tie 'la durtie d’immunisation i)ar voie intraveineuse sur le pouvoir 
floculant des serums, Compt. rend. Soc. biol. 143: 1478, 1949. 

P6rez, J. J., and Alazurek, C.: StTums de chevaux anti-proteides: Etude 
(juantitative de I’influence de la duree de I’immunisation par voie iiitra- 
veineuse sur le pouvoir precipitant des serums, Compt. rend. Soc. bio . 

144:1039, 1950. . 

n-effers II. P.: Heidelberger, AI., and Freund, J.: Antiiiroteins m horse 
sera: IV. Antibodies to rabbit seium globulin and their interaction 
with antigen, J. Exper. Aled. 8(): 95, 1947. . . 

'apiienheimer, A. AI., Jr.: Studies on diphtheria toxin and its reaction 
with antitoxin, J. Bact. 43: 273, 1942. • 

Irabar, P., and Staub, A. AI.: Etude (piantitative ties preciintations ob- 

74 . 


sorvpcs avoc certains extraits dii /?. nnthT(ici!< et un s6runi antichar- 
bonneux de clieval, Ann. Inst. Pasteur 71: 385, 1945. 

75. Kendall, F. F. Immiinochemistry: The quantitative relationship be¬ 

tween antigen and antibody in the jjrecipitin reaction, Ann. Now \ork 
.\cad. Sc. 43: 85, 1942. 

76. Heidelberger, M.; Rabat, K. A., and Mayer, M.: A further stiuly of the 

cross reaction between the specific polysaccharides of type III and 
VIII pneumococci in horse serum, J. Exj)er. Med. 75: 35, 1942. 

77. Pappenheimer, A. M., Jr.: Diphtheria toxin: II. The action of ketene 

and formaldehyde, J. Biol. Chem. 125: 201-208, 1938. 

78. Warner, R. C., and Weber, I.: The preparation of crystalline conalbu- 

min, J. Biol. Chem. 191: 173, 1951. 

79. Pope, C. C.; Stevens, M. F.; Caspaiy, E. A., and Fenton, E. L.; Some 

new observations on diphtheria toxin and antitoxin, Brit. J. Exper. 
Path. 32: 246-258, 1951. 

80. Topic}' and Wilson’s Principles of Bacleriology and Inununify (3d ed.; 

Baltimore: Williams Wilkins Cb., lt)46). 

81. Jerne, N. K.: A study of avidity, Acta path, et microbiol. Scandinav. 

Supp. LXXXVH, 1951. 

82. Bowen, H. E.: The Iiomogeneity of purified dii)htheria toxins and toxoids 

as investigated by the semisolid precipitin technic, J. Immunol. 
68: 42!)-439, 1952. 

83. Ouchterlony, O.: Antigen-antibodv reactions in gels, Arkiv Kemi, Mineral. 

Geol. 26B: 1-9, 1948. 

84. Agner, K.: Studies on peroxidative detoxification of purified dijditheria 

toxin, J. Exper. Med. 92: 337-347, 1950. 

85. Pillemer, L.; Toll, D., and Badger, J. S.: The isolation and characteri¬ 

zation of diphtherial toxoid, J. Biol. Chem. 170: 571-585, 1947. 

B. Specific Precipitation in Gels and Its 
Application to Immunochemical Analysis* 

J. OUDIN, Institut Pasteur, Paris 

Bechhold in 1905 (1) studied the reaction between antigens and 
antibodies in gels, reactants whose property of forming a precipitate 
had been known since the work of Kraiiss in 1897. While working 
with goat serum and its antibodies, the former observed the forma¬ 
tion of ring structures in his tubes which resembled Lieseganjr 
nnp, a phenomenon first oliserveil in 1890 with mineral reactants 
Reiner and Kopp in 1927 (30) also described what they lielieved 
M as Liesegang ring formation during specific precipitation 

fn. media has been used 

toi the identification of pneumoccjccus types in which an onumm 
nnsye^^ped around colonies srown on layers of agar contili'ning 



the homologous antiserum. At almost the same time Petrie (34) 
made more detailed observations when he attempted the trans¬ 
formation of '‘smooth” into “rough” bacterial types by culturing 
them on media containing antiserum. The specific halos which 
formed around meningococcus, pneumococcus and Shigella 
dysenteriae Shiga colonies often had a uniform appearance at the 
start, only to assume at a later stage the form of Liesegang rings; 
a similar type of precipitation was observed when solutions of the 
corresponding polysaccharides diffused into the same gel. Alae- 
graith (14), Kirkbride and Cohen (13) and Petrie and Steabben 
(35) applied similar techniciues to the study of meningococcus 
and to the identification of gas gangrene anaerobes, t These investi¬ 
gations accomplished their immediate bacteriologic purpose, and 
the occasional observation of concentric rings was only an accessory 
residt vaguely classified as the Liesegang phenomenon or mentioned 
without attempt at interpretation. The interpretation of con¬ 
centric ring formation as the manifestations of several antigens 
was not considered, t 

The qualitative analysis of a naturally occurring mixture of 
antigens presents a problem whose resolution has always necessi¬ 
tated a combination of physicochemical methods, such as salting 
out, with the methods of immunochemistry and antibody absorp¬ 
tion. Fractionation has proved necessary, since the classic immuno¬ 
chemical methods (those of Ramon, of Dean and Weber and of 
Heidelberger and Kendall) are essentially methods for titration 
of a given antigen or antil)ody and may be most successfully applied 
under conditions where only a single antigen-antibody reaction 

obtains. i ■ i 

However, molecules (part icularly protein molecules) endowed with 

the same immunologic specificity are generally better characterized 
by this specificity than are molecules having the same solubility 
in a given solvent or any other common physical property. More¬ 
over the immunologic specificity of proteins is related to then- 
precise biologic origin. It should therefore be possible to effect the 
analysis of naturally occurring mixtures of antigens without re¬ 
course to physical methods. 1 • I 1 -v’ 

In this chapter we shall use the term “immunochemical analysis 
to denote the determination without resort to previous physico¬ 
chemical fractionation of the number of antigens present in a 

KsTOT'Thrrh^fhmk preoh.itatioi. ot tl.c pncinnococcu, polysaccharides with ai.ti- 
pneumococcus serum (4). 



mixlure and tlioir identification. In seeking a solution to the prob¬ 
lem thus defined, it seemed to us that if the effect of gravity could 
be eliminated in carrying out the precipitation reaction at an 
interface, each antigen-antibody system present might give rise to a 
specific zone of precipitation. This hypothesis was verified by incor¬ 
poration of one of the specific reactants in a gel. The 1st results 
using this technique were published in 1946 (27, 28) and showed 
that under appropriate conditions: (1) a single antigen gave rise 
to only 1 zone of precipitation, and (2) when 2 distinct antigens were 
present in the same solution, each behaved independently as 
though the other were not present. 

Thus specific precipitation during the course of diffusion through 
gels constitutes the basis for a general method of immunochemical 
analysis. The techniques which have been described are conven¬ 
iently classified as techniques of simple diffusion and of double 

1. In our 1st paper (27) a technique for simple diffusion in 1 
dimension (in tubes)§ was briefly described, discussed and applied 
to the analysis of antigens in horse senim, and double diffusion 
(in tubes) was suggested as a possibility. 

2. Techniques of double diffusion in 2 dimensions (using agar 
plates) were developed by Ouchteiiony (19) and by Elek (7) to 
study the toxin production by bacteria. Ouchterlony studied, in 
various ways, the precipitation reaction in gels between diffusing 
substances, derived either from whole cultures or from toxin-con- 
tammg filtrates, and immune sera. Elek, on the other hand, cul¬ 
tured different strains of Corynchacterium diphtheriae and of 
staphylococcus in agar plates through which the constituents of 
the antiserum diffused. Both workers observed multiple zones of 
precipitation which they considered might have been caused by 
a multiplicity of antigens. Subsequently they applied their re- 

analysis of mixtures of antigens. Since 
1946, the Liesegang phenomenon does not seem to have been men¬ 
tioned 111 connection with antigen-antibody reactions in gels 


1 he following discussion is concerned with the 2 phenomena 
rroStgel. during tlie course of diffusion 

diffusion (6, 11).—If a given solution and the corresDondimr 
pm-e_s^vent are layered 1 over the other without mixing in f 

has been applied b^thL^^and a clrtafn^nmnlDer technique, which 




vertical tube, molecular agitation will result in penetration of 
molecules of solute into the solvent layer. 

M hen the concentration of solute is sufficiently low so that 
intermolecular action is negligible, the process is described in a 
general way by Fick’s law which relates the concentration c of 
the diffusing substance at any level x and time t 


where D is the diffusion coefficient of the diffusing substance. 

Integration of this equation leads to the following: 

1. When diffusion takes place from a solution of constant con¬ 
centration, Stefan has established a relationship between the in¬ 
itial concentration Cu and the concentration c at time / of the diffus¬ 
ing substance of diffusion coefficient I) at a point in the column of 
solvent at distance x from the interface of a solution-solvent 


where y 


2. In the case of a column of infinite length on both sides of the 
initial interface between solution and solvent 


Equation (3) differs from equation (2) only by the coefficient 
V /2 which affects the right-hand side of the eciuation. The equations 
are represented by curves shown in Figures 1, 18, 21 and 22. 

In the case of equation (3), the concentration at the level of the 

initial interface is constant and ecpial to ^ for all values of / > 0. 

It follows from the preceding eciuations that tor any concentration 

c, < Co in eciuation (2) and ^ in eciuation (3), at a point x where 

the concentration will always be e(iual to c,, one obtains foi all 
values of t 



= r‘ 

(■ 1 ) 

It should he added at lids point that coefficients of dilTusion in 
crease with temperature. 



SPECIFIC precipitation''. —Tlip different precipitating systems 
can be classed as follows: (1) Simple precipitating systems in which 
a single antigen reacts with an antibody. In this case we are not 
concerned Avith the presence or absence of other antigens or of 
other antibodies which do not enter into the reaction or with 
whether or not the reaction is homologous. (2) Complex precipitat¬ 
ing systems in which several antigens cross-react with a given 
antibody, e.g., the reactions between the ovalbumins of chicken 
and duck with anti-chicken ovalbumin serum. (3) Multiple pre¬ 
cipitating systems consisting of antigens anil antibodies reacting 
simultaneously which may be components of several simple or 
complex precipitating systems. 

There exists for each of the 2 reactants, antigen and antibody, 
a minimum concentration below which precipitation is not visible. 
As discussed by Cohn in the preceding chapter, the composition 
of any precipitate is determined by the ratio of the concentrations 
of antigen to antibody in the mi.xture, and there exists a ratio of 
equivalence at which both reactants are completely precipitated. 
The specific complex is soluble in excess of antigen; in fact, one 
can take a washed specific precipitate lormed at the equivalence 
point and dissolve it more or less completely in excess antigen. In 
the case of the flocculation reaction (^fliorse'^ f'yp^)> the specific 
complex is also soluble in excess antibody, in contrast to the pre¬ 
cipitin reaction (“rabbit” type). Because we will have occasion to 
relei to this later, it should be recalled that in both cases maximum 
total precipitation occurs at the equivalence ratio or in a slight ex¬ 
cess ot antigen when antibody remains constant (pp. 314 ff.). 

The reaction of precipitation occurs in 2 steps: (1) combination 

of antigen and antibody, and (2) aggregation leading to a pre¬ 
cipitation. * 

The time for the 1st stage is very short and in 1 instance has been 
shown to be less than 3 sec (15). On the other hand, the 2d sta^e 
olten takes an appreciable length of time which varies, depending 
on the latio ot concentration of the reactants. The optimal pre¬ 
cipitation titrations ot Ramon and of Dean and Webb depend 
on the time differential m precipitation. For the flocculation 
reaction, the rapid visible precipitation generally occtl a 
a point m the middle of the equivalence zone, and for tlm preOnith 
reaction at or near the equivalence ratio. 


Bnder this imperfect but convenient title wo will inni i 
cases m which 1 of the reactants i o f ^ ^ include those 

leactants, ,.e., antigen or antibody, diffuses 

S« Cohn, Reaction in Liquid Media,” this volume. 


into a gel containing, at the beginning of the experiment, the same 
concentration of the other reactant, i.e., antibody or antigen, at 
every point in the gel. The latter reactant will be called “internal” 
reactant and the former, the “external” or “diffusing” reactant, 
according to the terminology adopted by others who have used 
analogous techniques with inorganic substances (38). 

Agar has been almost the only substance used for preparation 
of gels. Gelatin has limited usefulness because the temperature 
at which it liquefies is too low. 


Technique. —Preparation of agar: A given weight of the purest 
agar obtainable is cut very fine and washed several times with 
distilled water, after which the pH should be close to neutrality. 
The agar is suspended in distilled water in 4-5 per cent concentra¬ 
tion by weight and placed in a boiling water bath until dissolved, 
then autoclaved for 15 min at 115 C. The solution is clarified while 
hot either by centrifuging or, more simply, by filtration through 
double sheets of fluted coarse filter paper.* The agar should he 
perfectly limpid while in the liquid state. 

The concentration of agar always decreases during this pro¬ 
cedure. Therefore the dry weight is determined and distilled water 
added so as to obtain, for example, a concentration of 2.4 per cent— 
a percentage high enough to allow for dilution with the internal 
reactant. Solid NaCl is added to final concentration of 0.85 per 


For future preseiwation and use, known volumes of the agar 
are distributed in glass tubes which are sealed hy flame and steii- 
lized twice at 1-day intervals by heating at 100 C for 30 mm in an 
open autoclave. 

Preparation of tubes: The width of tube chosen depends to a 
certain extent on the amount of material at one’s disposal. Ihere 
is no reason to choose a tube of greater than 4 mm I.D. One can, 
with a bit of experience, use tubes of 1.5 mm I.D In our quantita¬ 
tive studies (31,32) we have used tubes of 3 mm I.D. If one vishes 
lo make miciophetometric recordings, it is convenient to use 
cells with parallel walls, e.g., having 1.5 X 5 mm cross-section 
(cells with these dimensions were used in the reactions illustiated 

in Figs. 2-4, 6, 7, 9, 15 and 16). 

Normally, the agar will not stick to glass, so to prevent the ex¬ 
ternal reactant from flowing into the space between gel an(l 1 
the tubes are treated in the following manner. They aie heated to 
60-70 C and filled with 0.6-1.0 per cent agar solution at the sam 

* In our laboratory, Chardin paper is used. 



temperature. They are immediately emptied and plunged into an 
ice bath; their inner surface thus becomes coated with agar, which 
is dried by placing the tubes in a vacuum desiccator over sulfuric 
acid or phosphoric anhydride for 2 or 3 days. This coating can be 
detected by its opalescence, which disappears as soon as a solution of 
agar is added. The coated tubes may be used for several months. 

In nearly every case we have used a final concentration of 0.3 
per cent agar in mixture with internal reactant. This low concentra¬ 
tion of agar results in a limpid wet gel which has the consistency 
required when narrow tubes are used (a higher concentration coidd 
naturally be chosen). For most work the agar solution at 0.6 per 
cent is diluted in equal proportions with internal reactant. However 
when minimal dilution is desired, the original 2.4 per cent agar 
solution may be mixed with as much as 7 vol of the internal 
reactant. It is important to add a conv'enient antiseptic, e.g., 
Merthiolate in concentration of 2 X 10“'^ by weight. 

The original agar solution is melted in boiling water, then placed 
in a bath at 46-48 C and mixed at this temperature with the in- 
teinal leactant, previously clarified with care, e.g., by centrifuga¬ 
tion, and brought to the same temperature. 

If the internal leattant is an antiserum to be used at different 
concentrations in several tubes, it is often preferable to have the 
total concentration of serum the same in all tubes; to do this, 
the antiserum is diluted in the serum of an unimmunized animal 
of the same species. For quantitative observations, this precaution 
is taken in order to have the concentration of nonspecific substances 
roughly the same m the gel (see p. 351). 

The reactants are distributed in the coated tubes with a Pasteur 
pipet whose delivery is controlled by a rubber tube to the mouth 
For the height of the column of gel, 35-45 mm is convenient 
So that the meniscus of the interface will be of relatively constant 
hape m all tubes, it is best to fill them to a level several mm higher 
than that finally desired and immediately withdraw the excess 
-Yter distribution of the agar mixttire (at 46-48 C) the tubes are 
placed at the temperature of the experiment. Once the gel has 
solidified the external reactant is added in as high a concentration as 
po,ssihle and at the same temperature to anmnvl„ , as 

height. A 10-20 mm height can he u.sed if nece s ™ hu 
IS preferable. It may be coiivenipni r 4 but 30 35 mm 
layers in gelified meX (', 351 T. ' ’ both 

The tubes may be sealed with rubber cans or wRn o • 4 

of wax and tar or with modeling clay etc to^nrevpn^ 


avoided (see p. 355). A simple way to avoid artefacts due to changes 
of temperature is to enclose the tubes in a small container (a box 
or glass jar) linetl with a heavy layer of cotton anti not to remove 
them until the moment chosen for their examination, e.g., at the 
end of 4 or 7 days. 

Tn this laltoratory, observations are habitually made within 1 
week after commencement of the experiment. When a rapid an.swer 
is necessary, the essential data can often be obtained in 1 or 2 days. 
The tubes may be read against a totally black background with 
side-lighting or against a uniformly lighted background. 

A simple and inexpensive system allows one to photograph as 
many as 13 tubes of 2 mm I.D. at 1 time w ith a 2-fold enlargement 
on a photographic plate 9 X 12 cm (Fig. 17). The lighting is fur¬ 
nished by 2 broad horizontal light sources placed between the tubes 
and the black background at ecjual and constant distances above 
anti below the axis of the objective. The sharper the angle formed 
between the line from light source to the object anti the axis of the 
camera objective, i.e., the farther the light sources are from the 
tubes, the more readily will zones of WTak precipitation or oi 
small particles be photographed (Fig. 9 could have been taken only 
in this way). The w'ider the angle, i.e., the closer the light sources 
are to the tubes, the more readily may boundaries situated close 
to one another be distinguished. 

For quantitative studies, it is necessary to measure penetration h 
(see p. 348). In this laboratory these measurements were made 
at first by means of an apparatus built specially for this purpose 
(32). Alore simply, and with apparently sufficient precision, 
Aliinoz and Becker (17) use calipers. The measurements are now 
made by the author from photographs on glass obtained directly 
with a 2X enlargement. 

Theoretical Discussion.— Dcnsf/y of preczpiiale will be dehnecl 
as the quantity of precipitate/unit vol, where unit vol denotes the 
smallest volume whose content or precipitate may be considered 
homogeneous. For lack of a simple and raiYd method ot measuring 
the density of precipitate, w'e must be satisfied with a comparison 
between different points in the same gel. It is assumed as proliable, 
within certain limits, that the point of maximum light scattering 
by the precipitate coincides with the point of maximum density 

The density of precipitate at any sivon |iomt in the gel is detei- 
minod liy a sei-ies of complex relationsliips winch are linked, on 

of precipitation does not appear illuminated. 



one luind, to phenomeiui of diffusion und, on the other, to immiino- 
ehemieal phenomena, not to mention (piantitative C'hanges due 
to the gelling substance itself and to particular properties of gels. 

Sintplifying Hypotheses .—The following simplifications allow 
us to describe api)roximately the phenomena on which the method 

A gelling substance is employed to avoid the effect of grav¬ 
ity and mechanical agitation which are unavoidable in lifiuid solu¬ 
tions. It is assumed that its jDresence will not bring about any 
other qualitative changes. In most cases the density of precipitate 
at each point in the gel is considered to be determined by the con¬ 
centration of the 2 reactants at the same points, in accordance 
with the laws Avhich determine (at the end of an appreciable time) 
the formation of a specific precipitate in liciuid medium using the 
same reactants at the same concent rations. J 

I'he foregoing assumptions neglect 2 facts: 

1. Tlie formation of a specific precipitate is not instantaneous; 
therefore, if the concentration of antil)ody at any given point 
varies but slightly and the concentration of antigen changes 
rapidly from 0 to a value high enough to inhibit precipitation 
(during a period too short for the precipitate to form), precipitation 
Avill never appear at this point ev^en though the reactants have 
doubtless combined. If one can consider the antigen-antibody 
combination as instantaneous, the absence of preci])itation under 
certain conditions (always close to the source of diffusion of the 
antigen in i)re(‘ipitin-type systems) does not appear to change 
the outcome of the experiment. INIore genei-ally, by neglecting the 
time of formation ol the precipitate, one overevaluates the time 
ic(iuiied for a given concentration ot the diffusing reactant to reach 
a given {wint in the gel, but ordinarily this error is negligible. To 
avoid this minor difficulty in our theoretical discussion the equiv¬ 
alence point will be referred to as the point or level in the gel where 
the concentrations of antigen and antibody are in their ratio of 
equivalence at the moment under consideration. It is clear that 
the equivalence point should lie close to the point of maximal 
density ol precipitate when the internal reactant is antibody if 
precipitation were instantaneous. ‘ ’ 

2 \\ hen the point ol equivalence migrates in the gel, anligen- 
antiboily complexes will be formed (depending on the point ton- 
si^demd^i^the gel) citlier by the direct combination oUlie 2 re- 

..r,,,, „f the k„„w„ 

confirmed satisfactorily liy experiment not nniV f ^ i ^ has been 

lusioii studied hut also tor doiihlc dittuslin (cf 

The L,ese,a,.g i,heuo.„c,.„„ will not be Xi'dtod S tI.L dliiii"^'' 


actants or by a new reaction of 1 or the other reactant with the 
complexes already formed. 

In conclusion, the simplifications lead us (I) to consider as instan¬ 
taneous the initial formation of the specific precipitate and its 
subsequent reactions, and (2) to suppose that the final density 
of precipitate is always the same for giv^en final concentrations of 
the 2 reactants. 

The case of simple diffusion will be treated as if the internal 
reactant itself does not diffuse when precipitation occurs in the gel. 
This is evidently not exact since molecular agitation, the cause 
of diffusion, involves the molecules of the internal reactant 
as well as those of the external reactant,§ but the resulting varia¬ 
tions in concentration of the internal reactant in the agar layer 
are slight compared with those of the external reactant, as has 
been verified by experiment (Fig. 4). 

It will be recalled that, under the conditions for applying equa¬ 
tions (2) and (3), the concentration of diffusing substance re¬ 
mains constant in the immediate neighborhood of the interface 
once diffusion has commenced. It will be assumed that this con¬ 
stancy is realized. Even though the experimental conditions de¬ 
viate from the ideal, certain predictions of the above theoretical 
equation may be verified satisfactorily by experiment (see p. 347 
and Fig. 5). It is quite possible that this confirmation is due partly 
to a compensation between the errors resulting from diffusion of the 
internal reactant and from changes in concentration of external 
reactant at the interface. 

Theoretical Analysis of Simple Diffusion .—The conditions for 
simple diffusion are apparently quite different from those for tree 
diffusion, above all when the ratio of initial concentration of the 
2 reactants approaches their equivalence ratio. In the partic¬ 
ular instance in which the ratio of initial concentration is 
that of equivalence, precipitation occurs in the immediate vicinity 
of the interface and neither of the 2 reactants penetrates to any 
appreciable extent into the layer containing the other reactant. 
On the other hand, when the concentration of internal reactant 
is very low with respect to the initial concentration of the external 
reactant the latter’s concentration at different levels in the gel 
will approach that expected in free diffusion, at least up to the 
point in the gel where the equivalence ratio of the 2 reactants 

becomes realized." 

"T^^^^t^rtefacts mentioned later are cau.sed in part by diffusion of the internal 



Between these 2 extremes (Fig. I) the concentration of external 
reactant at different levels in the gel cannot be determined with 
certainty from the laws of free diffusion. However, it follows 
from these laws that for a concentration of external reactant 
which remains constant at the level of the interface, and for a 
concentration of internal reactant which remains constant through¬ 
out the gel, the distance from the interface to the point of equiv¬ 
alence is proportional to the square root of time (equation (4)). 

Fig 1.— Diagram representing simple diffusion at a given time t, in conform¬ 
ity with simplifying hypotheses discussed in the text. Concentrations of e.x- 

of internal reactant C{in) are expres.sed in equivalent 
units, e.g., flocculating units of toxin and antitoxin; therefore equivalence 
point (F,) corre.sponds to the point of intersection of the 2 curves C(ex) curve 
I'.n r? w"'" r® theoretical equations of free diffusion, justiSle onTv 
when C{in) is neg igible with respect to initial values of C{ex) For this reason 

Moreover, if the concentration of tlie internal reactant remains 
constant at all levels of the gel, the density of precipitate 
at the equivalence level should also remain constani provided 

the time of formation of precipitate may be neglected * 

In practice, one nearly always chooses antibody as internal re 

r«7aX:.2i:d3)“ ^ 

Theoretical Basis for Determination of Number of i . • 

.cna-lt has been assumed that the conlr^L/ofT: ii^termd 


leactiint iGiiiciins consttiiit in tliG column ot gel. II ii single antigen 
is the diffusing reactant, the curve describing the density of pre¬ 
cipitate at a given moment as a function of distance from the 
interface will pass through 1 and only 1 maximum. 

this is to be expected, considering the known properties of the 
curve describing the amount of precipitate obtained as a function 
of antigen added to a constant amount of antibody, according to 
the quantitative method of Heidelberger et al. Aside from the simpli¬ 
fied assumptions introduced above for consideration of specific 
precipitation in tubes of gel, the only difference between reactions 
occurring in gels and those occurring in liquid solutions resides in 

Fig. 2 {left ).—Reaction of chicken ovalliumin (reactant diffu.sing from uiiper 
liquid layer) with homologous ralibit antiserum (lower gellified layer). Micro- 
photometric tracing was made from a photographic negative. Ordinates (dis¬ 
tances from each given level to the interface) are in the same scale for the 
tracing as for the photograjih of the cell in which the reaction took place. 
Abscissae are assumed to be proportional to 7’ = ///o (/ = light transmitted 
to photoelectric cell of microphotometer; h = light emitted). Interface is 
marked on the photograph by a white dash and recorded on top ot the tracing 
by a peak. The leading edge of the zone of precipitation is very sharp. The 
density of the precipitate passes through a maximum in the neighborhood ot 

this limit (29, 32). . . 

Fig. 3 {right ).—Antibody is the diffusing reactant (upper layer: anti- 
ovalbumin rabbit serum); antigen is in the gel in low concentration. Density ot 
precipitate increases from bottom to top until the interlace is reached, since 
the excess antibody does not dissolve the precipitate (29, 32). 

the fact that, in the former case, one does not know exactly what 
relationship describes the variation in the concentration of antigen 
as one passes from the point farthest from the interlace to the 
interface itself. However, it is known that this concentration 
increases in a continuous manner, and this fact alone leads to t le 
conclusion that in the case of a single antigen the density of pre¬ 
cipitate will never pass through more than 1 maximum. 1 here ore 
the number of antigens reacting in a system cannot be less than 
the number of observed maxima ot density of precipitate. 

Experimental Examples.-simplb precipit-vting systems 
The simple system that has been most tlioroughly studied is that 
of chicken ovalbumin and its homologous rabbit antiserum. 



1 . Wlieii oviilbumiii tliffuse.s into a gel containing homologous 
antibody (Pdg. 2) the zone of precipitation* has a sharp leading 
edge, while its upper part containing antigen-antibody complex 
partially or completely dissolved in excess antigen is diffuse. The 
density of precipitate passes through a maximum close to the 
leading edge. 

2. When rabbit antibody diffuses into a gel containing the antigen 
the density of precipitate increases without passing through a 
maximum anywhere between the lowest point of the zone and the 
interface, because of the absence of inhibition by excess of anti¬ 
body (Fig. 3). For this reason the choice of antigen as diffusing 
reactant is preferable for the purposes of immunochemical analysis. 


1 -' „ 




00 - J 




■ ■ ■ 

^=9J- 1 j. 

Fig. 4.— Development in time t of the zone of precipitation of chicken oval¬ 
bumin (e.xternal reactant) with homologous rabbit antiserum (internal react¬ 
ant), with corresponding microphotometric tracings (see legend, Fig. 2). 
hame initial concentrations of the 2 reactants in all 3 cells. Each white dash in¬ 
dicates level of the interface (29, 30, 32). j = days. 

Influence of the Principal Variables.—rime: The maximum 
density of precipitate changes only slightly during the time it 
takes for the leading edge to migrate from the interface into the 
gel (Fig. 4). The leading edge formed by precipitation of the diffus¬ 
ing reactant, ovalbumin, is very sharp. The leading edge is at 
hrst convex, like the interface, but after several hours tends to 
flatten out and form a horizontal plane. The displacement or 
penetration, h, of the leading edge can then be measured with a 
precision of 0.1 mni, and is a linear function of the square root of 
bTlt^ Pi-^icted by the theory; but h at 0 time, as determined 
aLm oT ‘ ""’if generally situated slightly above the interface, 

about 0,3 m m on the average in the experiments cited here (Fig. 5). 

''T.r'mCha.T ” 

constant (as Ptc<HMo<fbv™FiJk4“aw)^in'c“ pe^ V remained nearly 

umns or ,,uid, o, gelatin and or 


Concentration of the 2 reactants: Experiments designed to study 
the influence of concentration on the maximum density of pre¬ 
cipitate are illustrated in Figures G and 7. These figures demon¬ 
strate that the initial concentration of antigen (diffusing reactant) 
does not significantly influence the maximum density of pre¬ 
cipitate. The maximum density, on the other hand, varies consider¬ 
ably with the concentration of antibody (initial reactant). We 
would suspect a priori an approximate proportionality between the 

0 10 20 50 100 150 

1j. 2j. 3j 4j. 5j, 7j 9j 12j 15j. 

Fig. 5.—Measured distances (mm) of boundary of precipitation of oval¬ 
bumin from an arbitrary starting point plotted as function of the square root 
of time t (min) for 6 tubes. Each horizontal dash along the ordinate h indicates 
the level of the interface in each tube. Initial concentration of the internal re¬ 
actant (antiserum) was equal in all tubes; figures on the right are logarithms 
of initial concentrations of antigen (32). 

maximum density and the concentration of internal reactant, 
provided the smallest concentration of antibody which gives a 
visible zone is negligible compared with the amount of antibody 
actually reacting. 

The influence of antigen and antibody concentrations on the 
penetration, h, is expressed in approximate fashion by the following 
relationships which have been established empirically for the case 

of a single antigen (31,32). ■ c 

1 . For each given value of the initial concentration a of the 


h . g 


where h is the distance from the interface to the leading edge, t 


the time of diffusion, g the initial concentration of the antigen, go 
the extrapolated value of g for which h/\/t = 0 , and 7 is a coeffi¬ 
cient > 0 . 

Fig. 6 {above ).—Photograph with microphotometric tracing (see legend, 
Fig. 2) showing influence of initial concentration of external reactant (oval¬ 
bumin) on zone of precipitation. Same reactants as in Figure 4; initial concen¬ 
tration of immune serum is equal in all 3 cells. Initial concentration of oval¬ 
bumin was proportional to values of g. Time was calculated from equation (5) 
to give approximately the same penetration in all 3 cells at the moment of 
photographing. Each white dash indicates level of the interface (29 30 32) 

I -—Photograph with microphotometric tracing (see legend 

I ig. 2) showing influence of initial concentration of antibody (internal reactant) 
on zone of precipitation. Same reactants as in Figures 4 and 6, Initial concen¬ 
tration of antigen was equal in all 3 cells; initial concentration of antibody was 
proportional to values of a Time was calculated from equation (6) for penetra¬ 
tion to be appro.ximately the same in all 3 cells at the moment of photograph¬ 
ing. Each white dash indicates level of the interface (29, 30, 32). ^ ^ 

2. For each given value of g 

h a 

/- = a log — 
Vt Oo 

( 6 ) 

Where a is the initial concentration of antibody, «« the extrapolated 
value of a for which h/y/t = 0, and a is a coefficient < 0 The 

a depend on the nature of the antigen. h/Wi 
direction as g and inversely as a. 

values of 7 and 
varies in the same 


By vnrying the concentration of only 1 reactant at a time, 
one can represent the variations of h/\/t as a function of log g 
and log a by 2 series of straight lines vhich, for the experiment 
cited here, are reproduced in Figure 8. The extrapolation of these 
lines to 0 for h/y/t gives ^udues for g and a Avhose ratio, except 
in extreme cases, is approximately that of the ecpiivalence ratio. 
As will be seen later, equations (5) and (6) cease to hold at high 
values of h/y/1. 

Coefficients of diffusion: Mitchison and Spicer (IG) attempted 
to apply the theoretical equations to diffusion of antibiotics in tubes 
of gel. Their calculations, made by means of expansion into series, 
showed that: (1) in equation (3), for low values, x/y/t tends to 

Fig. 8 . —Precipitation of ovalbumin in a gel containing immune serum. 
Figure shows variations in h/_\/t (h in mm; t in min) as a function of log ^ and 
log a, where g and a are initial concentrations of ovalbumin and anti-oval¬ 
bumin (31, 32). 

become a linear function of log c and the expression approaches the 
empiric equation (5), and (2) when x/y/i i« large, x^/t tends to 
become a linear function of log c. 

The 2(1 observation, which was verified experimentally in the 
case of diffusion of antibiotics, has been confirmed by Becker ct al. 
(2, 3) for the diffusion of antigens by using the present techiiuiue. 
wWn g is very high and a very low, they have shown that the slope 
of the line which represents log g as a function of h^/t is inversely 
proportional to the coefficient of diffusion of the antigen; this 

slope is equal to -b log e. They have lliiis dcvelopetl a metliod of 

determining D in rvhicli it is not necessary to use a simple system 
nor to know the concentration of antigen or antihoily. Jhe \alues 
for I) obtained hy this method for several allmmms agreed satis- 



factorily with the accepted values as determined by free diffusion 
(although slightly lower). 

The same authors indicated a 2d method of determining D when 
g is large and a is small; in other words, the condition which most 
closely resembles free diffusion. Starting with the hypothesis 
that at the level h of the leading edge of the zone of precipitation 
the ratio of concentrations of the 2 reactants has the same value 
R as in a mixture at equivalence, and granting that the concentra¬ 
tion of antibody remains equal to a for all values of h, they sub¬ 
stituted in equation (2) the known value of g for Co, the value of h 
measured at time t for .r and the known value of Ra for c. R is ob¬ 
tained by the classic immunochemical method which requires 
an antigen of sufficiently high purity, contrary to the preceding 
means of determination. The values of D found by this means were 
close to the preceding. 

Other Variables: Finally, the influence of other variables should 
l)e mentioned briefly. The value of h/\/t increases at higher tem¬ 
peratures. If the antigen is not present in pure solution, the value 
of h/y/l may be increased due to the presence of nonspecific sub¬ 
stances provided their concentration is sufficiently high.J On the 
other hand, as the concentration of diffusible sid)stances other than 

the antibody in the gel is increased above a certain value, a diminu¬ 
tion of h/y/l results. An increase in concentration of agar will 
also decrease the value of h/y/l. 

From the foregoing discussion the limiting conditions for the 
appearance of a zone in the gel may be stated as follows: (1) The 
concentration of antibody must be sufficiently great for the zone 
of precipitation to be perceptible. (2) The‘ratio of the initial 
concentration of antigen to the concentration of antibody must 
be distinctly greater than their equivalence ratio (see p. 350). 

A fairly larg^e number of simple precipitin-type systems has 
leen studied. I hey have given a single zone of precipitation with 
a single maximum of density of precipitate. 

Most protein antigens show a zone of precipitation with a sharp 
leading edge. In 2 cases investigated, polysaccharide haptenes gave 
a zone with a diffuse leading edge (Fig. 9). ^ 

Figures 10 and 11 show the type of reaction obtained with 
flocculahng antibody (horse diphtheria antitoxin). When the 

^ soU.tions of t„e same 


external reactant in the flocculating system is the antigen (Fig. 10) 
the result is similar to that obtained between precipitin-type 
antibody and a protein antigen. On the other hand, when \he 
external reactant is flocculating antibody (Fig. 11), the result is 
very different from that of precipitin antibody (Fig. 3), due to solu¬ 
bility of the specific complex in excess flocculating antibody. It 
may be con\ enient, particularly in such cases as the present one, 
to incorporate both reactants in layers of gel, so that the zone 

Fig. 9 (left ).—Photograph and microphotometric tracing (see legend, Fig. 
2) of zone of precipitation of the O polysaccharide of Eberthella typhosa (ex¬ 
ternal reactant) with an antisomatic serum. This figure contrasts with those of 
the reaction of ovalbumin by absence of a sharp boundary (29, 32). 

Fig, 10 {center ).—Zones of precipitation of purified diphtheria toxoid (e.x- 
ternal reactant) with horse antitoxin (internal reactant) at varying initial 
concentrations of serum in proportions of 1, Vs, Vio. Appearance of the 
zones resembles those of Figure 2 with rabbit antiserum as internal reactant 
(33). The toxoid used in experiments for Figures 10, 11 and 19, D, was a highly 
purified product prepared by Dr. L. Pillemer; its activity, determined by Dr. 
Melvin Cohn, was 2,150 Lf/mg N. The antitoxin, supplied by Dr. LeiVRtayer, 
contained 1,600 units/cc. 

Fig. 11 {right ).—Zones of precipitation of horse diphtheria antitoxin (same 
antiserum as in Fig. 10, but used undiluted as external reactant) with purified 
diphtheria toxoid (internal reactant) at varying concentrations of toxoid in 
proportions of 1, V 2 , V 4 - Appearance of the zone is very different from those 
in Figure 3 because of solubility of the precipitate in excess antibody (33). 

of precipitation may be observed in either layer, depending on 
the ratio of the initial concentrations of antigen and antibody, 

COMPLEX PRECiPiT.\TiNG SYSTEMS. —When 2 01’ more antigens 
in a given system cross-react, i.e., when part of the antibody is 
capable of being precipitated by 1 or the other of the antigens, 
they cannot react independently of each other. In these ciicum- 
stances the penetration h of each leading edge is not only a function 
of the concentration of the particular antigen but also a function 
of the concentrations of the other cross-reacting antigens (Fig. 12). 

If 2 or more antigens cross-react with the given antibody in 
a gel to give several zones of precipitation, the zone farthest 



from the interface is due to the antigen which precipitates the 

least amount ot antibody (P ig. 13). 

MULTIPLE PRECIPITATING SYSTEMS. —Ill analyses of naturally 
occurring mixtures of antigens, multiple precipitating systems will 
usually be obtained. In such a system all antigens which do not 

.A A *i i i 

9p/2.S = 1 



1 0 

1 0 


9c/2.6 = 1 

I 1 

1 1 

1 V 4 

V 4 

V20 V20 




9»/2.5 = 11111111111 

9 «/ 2.5 = 1 1 V2 V4 V10 '/ 2 Q V40 V100 ^/200 'AOO 0 

Pig. 12 (a6oi’e).—Study of a complex system. Reaction of anti-chicken oval- 
bumin rabbit serum with mixtures of chicken and duck ovalbumin and with 
each alone, gp and go are concentrations of chicken and duck ovalbumin re- 
spectively. Penetration of the zone of duck ovalbumin is clearly increased by 
presence of chicken ovalbumin in 2d and 4th tubes, l)v comparison with 3d 
and 5th tubes. Reverse influence is more subtle. E under the 1st tube indicates 
that the antiserum had been absorbed with duck ovalbumin (30, 32) 

d>eloxo).-~^An\G complex system as Figure 12 (see legend). Constant 

ovSbuminTrnm^ut^^^"^ decreasing concentrations of duck 

9 i ^7 i ^ certain concentration of duck ovalbumin 

?re2ed ^3% penetration of the chicken ovalbumin zone is in¬ 

cross-react will behave independently, i.e., as if each were presem 
alone.^ Figure 14 shows the behavior of 1 complex and 2 simple 
systems an d that of an artificial mixture containing all 3 systems 


under comparal)le coiulitioiiK. It is dear that the picture of the 
multiple system corresponds closely to that which would be ob¬ 
tained on superimposinfr the photographs of the isolated systems. 
Applications to Immunochemical Analysis.—determin.4 - 

OF THE TECHNIQUE.— We may conclude from the foregoing dis¬ 
cussion, with certain reservations considered later (p. 355), that the 
number of antigens in a given solution is greater than or equal to 
the number of leading edges or of maxima of density of precipitate 
observed in the gel. 

As already pointed out, the distance from each leading edge to 
the interface is nearly proportional to the square foot of time. The 

Fig. 14. —Study of an artificial multiple system. Internal reactant consists 
of the same mixture of antisera in all 4 cells. Antigens are: in A, polysaccharide 
of E. typhosa; in B, a mixture of chicken and duck ovalbumin; in C, horse serum 
albumin; in L), mixture of all 4 antigens, each at the same concentration as in 
A, B and C. 1) looks like a superposition of the other 3 (30, 32). 

ratio of distances between different leading edges, therefore, re¬ 
mains constant with time. 

The different zones are at first (if precipitation is rapid) super¬ 
imposed on each other in a very narrow space just below the inter¬ 
face. In the course of time they become progressively more dis¬ 
tinct as if they had been magnified, depending on their number, 
density and the value of h/\^i for each zone of precipitation. It 
remains possible, of course, that 2 leading edges may be so close 
together that they cannot be distinguished or that a very faint 
zone may be masked by 1 ot high density and greater penetiation. 

The determination of the number ol antigens depends mainly 
on the concentration of their corresponding antibodies in the 
serum used and, to a lesser extent, on the concentrations of the 

antigens themselves. _ 

1 . For each antibody there exists a coiu'cntration below wlucli 

precipitation is not visible. The sensitivity in gel is ot the same 

order as in liquid media, or slightly less. 



2. The miiiiniuni concent nit i()n ot ejich antiij;en thiit cun ie<A(t 
to form u visible zone depends on the concentration of antibody 
and on the antigen-antibody ratio at equivalence. If the concen¬ 
tration of antigen is less than the minimum value reiiuired for 
equivalence, precipitation occurs in the antigen layer. Lor this rea¬ 
son it is advisable that the concentration be as high as possible. 
A high concentration does not introduce any complications. If one 
wishes to be sure that no antigen present in low concentration will 
escape detection, it is sufficient to use serial dilutions (e.g., 2- or 3- 
fold) of the antisei'um in the gel. Certain zones disappear as the 
concentration of certain antibodies becomes too low in the higher 

Fig. 15.—Photographs of striations or narrow bands jirovoked liy rapid 
changes in temperature. Same cells, 1 and 2, were photographed successively 
in A and B. After being allowed to develop for several days at 22 C, the cells 
were placed, one at 37 C {1: increased density of striation), the other at 4 C 
{2: reduced density of striation). A was taken at the moment the 2 cells were 
replaced at 22 C'. This change caused, at the level of the boundary, a new stria¬ 
tion, seen in B taken several days later. C, photograph at greater magnifica¬ 
tion of 2 other cells: 3, placed at 37 C for about 15 min, then returned to 22 C; 
4, placed for 15 min at 4 C, then returned to 22 G (30, 32). 

dilutions of antiserum (see (I)). The remaining zones are easily 
recognized trom 1 tube to the other, and the new zones tvhich ap¬ 
pear on dilution of antiserum may be recognized by their lower 

values of/i \/b 

Sources of Enor .~]. Nonspecific precipitation: The pi'oblem is the 
same here as in reactions in liquid media and is handled similarly. 
Normal serum controls described previously should be set tin in 
the gel (see pp. 301 ff.). ' ^ 

2 . Artefacts due to ehanges in temperature: A i-apid change in 
temperature, even small, can provoke the formation of a narrow 
band whose density is different from that of the surrounding pre¬ 
cipitate (Fig. 15). These artefacts appear at the level of the 
leading edge at the time of the temperature change and remain 
hxed at th is level while the zone continues to migrate. They are 

th! IS possible that the appearance of the tubes observed bv Brown u 

ttej-eact.o„ was allowed to proceed in the refrigerator, wile tTartelaiis oT this 


due t o a .sluirp change in the progres.s of the diffusion and demon¬ 
strate that the dittusion of antii)ody is not 0, even though, as a 
simplitying assumption, it has been neglected. The appearance of 
these artefacts is rather characteristic; moreover, their lack of mo¬ 
bility contrasts sharply with the regular progression of the true 

3. The Liesegang 'phenomenon should be mentioned here in view 
of the role attributed to it in the past. With inorganic substances 
such as K 2 Cr 207 and AgNOs, it consists essentially of a discontinu¬ 
ous precipitation in the form of separate bands whose thickness and 
spacing are periodic (11, 39). Such successive bands remain immo¬ 
bile (39); in our studies we have never observed this phenomenon 
in immunochemical reactants. 

problem of identification is the following: 

Fig. 16.—Photograph of 2 cells each containing 
3 diffusing antigens in the same gel. The only 
difference between them is that the concentration 
of 1 antigen, ovalbumin, is much higher in B than 
in A. Penetration of the zone of this antigen is 
therefore greatly increased in B as compared with 
A (30, 32). 

A given solution M, containing several anti¬ 
gens, reacts with an anti-d/ serum A^ to 
give several zones. One wishes to know which of these zones are 
produced by antigens common to M and to another solution L, 
suspected to contain 1 or more of them. 

When the antigens in M and L capable of reacting with are 
few in number, e.g., 3 or 4 in AI and 1 or 2 in L, one can, in fa\ oi- 
able circumstances, hazard a guess as to the coriesponding zones 
by comparing tubes in which AI and L react sepaiately vith A^f. 
Any identical antigen present in both tubes will give a zone of the 
same maximum of density of precipitate, a property which is inde¬ 
pendent of concentration. 

Idle independence of reaction of different antigens suggests sev¬ 
eral procedures for identifying them with certainty, for if one 
changes the concentration of only 1 antigen or of its antibody with¬ 
out affecting the concentration of the other reactants (antigens or 
antibodies), only 1 zone will be affected (this would not necessarily 
be true in the special rare case in which one changed the concen¬ 
tration of 1 of 2 cross-reacting antigens mM). 

1 . The 1st consists in comparing 2 tubes, I of which contains 



the antigens of L in higher eoneentration than the other (Fig. 16). 
In practice one prepares a dilution of I\f in Ij and a similai’ dilution 
of M in any convenient diluent (normal saline, or serum Irom un¬ 
immunized animals). Each dilution is allowed to react separately 
with Am in tubes. If comparison reveals that certain zones in M -f 
L show a higher penetration than in tul)e M, these zones are caused 
by antigens present in L. The higher the concentrations in solu¬ 
tions L of the antigens common to both M and L, the more suc¬ 
cessful will this procedure be, provided the zones due to the reac¬ 
tion M-Am are not too numerous to make comparison between the 
tubes difficult. 

2. The author usually prefers to use another procedure which 
involves the more or less complete absorption of the antibodies to 
L which are present in A^. Three situations are possible: 

a) Only a single antigen is common to M and L, which 
allows the determination of the ecjuivalence ratio of L-A^ and the 
absorption of A^ by a slight excess of antigen L. 

b) Several antigens are common to M and L, and it is both use¬ 
ful and easy to absorb A^ completely with L. One carries out this 
absorption by mixing A^ with a quantity of L such that the super¬ 
natant no longer precipitates with L. By comparison of tubes in 
which M reacts respectively Avith this supernatant and with unab¬ 
sorbed Am (similarly diluted), the zones common to M and L may 
be identified since they di.sappear after absorption; the zones which 
remain correspond to antigens of M which are not present in L (see 
example in (32)). 

c) Several antigens are common to d/ and L, but only those 
found in L in an appreciable concentration are of particular in¬ 
terest. This is the case when M is separated into fractions, Li, L., 
etc., each containing all the antigens of M at least in trace’quanti- 
ties. Am is absorbed by a given quantity of a given fraction of L. 
All the zones which disappear after absorption are due to anti¬ 
gens present in L, but any zones which may remain can be due to 
antigens present in L in low concentration. 

The procedure involving absorption, although somewhat longer 
than procedure 1, is more certain. In contrast to procedure 1 and 
also to the supplementary means offered by the technique of 2- 
dimensional diffusion (coalescence of zones, see p 370) a great 
number of zones does not affect the ease of reading the tubes (Fig 

In the foregoing discussion we have considered the most general 
case namely, that of icIentifyinK zones due to antigens of identiea 
spe,Mfic'ty present n. both L and M using an antiserum homologous 
for M. Different results would, of course, he ohtained if certafn of 


the antigens of L and ^[ cross-reacted in anti-d/ serum. The in¬ 
vestigation of complex systems suggests that in tlie case of a 
cross-reacting antigen, procedure I, i.e., increase of antigen concen¬ 
tration, could result in (a) a slight increase in the penetration of 

17 —Identification by means of absorption. A human serum globulin 
fraction has been separatecTinto 9 subtractions by salting out: A, B, C, etc. 
In the 1st tube: reaction of the original fraction of antigens with homologous 
antiserum (8 zones). In the 9 following tubes: reactions of the original trac¬ 
tion with tile same immune serum absorbed by a constant quantity of each of 
the 9 fractions, respectively; every zone which disappeared (in comi>arison 
with the 1st tube) corresponds to an antigen present in the subfraction under 
consideration. This iiermits establishment of a scheme showing distribution ot 
7 antigens in the 9 fractions. The low penetration of zone 8 shows that con¬ 
centration of antigen 8 is weak. Wiien larger quantities ot fniction C mul D 
were used in the absoi'iition, zone 8 was moditii'd. By varying the amount used 
for absorption, one can construct a curve showing relative concentrations ot 
each antigen in the various fractions (33). 

the corresponding cross-reacting zone of M with a diminution in 
the maximum density of precipitate, and (b) the appearance of a 
new zone which migrates more rapidly than the homologous zone 
(Fig. 13 tubes right to left except the 1st tube on the left). 

Procedure 2 (absorption) would not show the complete dis¬ 
appearance of the corresponding zone, but only a diminution in its 



maximum density of precipitate and an increase in its penetration. 
Eotli effects become more marked as the absorption l)ecomes more 
complete, in other woi’ds, as the reaction approaches that of the 
homologous antigen (Fig. 13, compare 1st and last tubes). When 
thei’e are more antigens reacting, the analysis becomes more diffi¬ 


Simple diffusion in 2 dimensions, i.e., diffusion in a layer of agar 
rather than in a tube, is little used and has not been studied quan¬ 
titatively. A technique has been l)riefly described by Ouchterlony 
(19, 20, 26), who used diphtheria toxin diffu.sing from a “penicillin 
cup” into a layer of agar containing antitoxin. One or several 
zones of precipitation concentric at the source of the diffusing re¬ 
actant appeared in the gel after a variable length of time. Reasons 
for the delay are discussed on page 343. 

Although qualitatively analogous, it is incorrect to reason that 
the quantitative relationships which describe diffusion in 1 dimen¬ 
sion ai)ply here along each line which radiates from the center of 
the reservoir. Actually, in the tube, when the leading edge of a zone 
is at a distance h from the interface, the molecules which have 
crossed the interface redistribute themselves, unequally according 
to the laws ot diffusion, in a A'olume i)roportional to h. In two-di¬ 
mensional diffusion, they are I’edisti’ibuted in a volume propor¬ 
tional to Tr/i^. 

Petrie observed the coalescence of haloes which developed around 
neighboiing bacterial colonies (34). It will be seen later hoAv the 
coalescence of zones of precipitation may serve as a means of iden- 
tihcation m tlouble diffusion. No one appears to have used the coa- 
escence of zones produced by the same antigen, diffusing from 2 
different sources in a gvl containing immune serum, for purposes 
ot identification. 


^?i diffusion, 2 reactants diffuse to- 

vaid each other through a gel which contained neither reactant at 
the begmnmg of the experiment. The pcsibility of douhirdiff," ion 
in 1 (hmen.',ion (in tubes) as a method of immimoehemical analysis 

las siiggest^ed early by the author (27). However, no precise tech- 
nique has been published.* ^ 

( 35 a) appeared i„ 

of anthseruin incorporated in gelatin is ninr!!?) ■ +i ^ ^sed. A laver 

covered with a layer of gelatiT ^ bottom of each tube. Thi^s 

gelatin The diffusion is carried out in the cold^ The %t[ antigen in 

unpublished work by Oakley. ^ ‘Authors credit the technique to 



Several procedures for double diffusion in 2 dimensions (in layers 
of agar) have been descril)ed by Ouchterlony and by Elek. In their 
early papers (19, 8 ) both described techniques designed to study 
toxin production by living cultures, in particular C. diphtheriae 
(see p. 374). Later they described other techniques applicable to 
the analysis of solutions of antigens (22, 25, 8 ). 

In the present discussion, “source of diffusion” means the bound¬ 
ary which separates the reservoir containing a solution of the 
diffusing reactant from the gel into which the diffusion takes place. 
Two cases will be considered in which the sources of diffusion are 
rectilinear: ( 1 ) diffusion of antigen and antibody from parallel 
sources which face each other, forming opposite sides of a rec¬ 
tangle (see Fig. 20, A.); (2) diffusion of antigen and antibody from 
sources perpendicular to each other (see Fig. 20, B and C). 

We will limit the discussion to these 2 cases, although any number 
of other techniques could be imagined, varying according to the 
form, number and position of the sources of reactants. What is 
true for the 1 st of these 2 techniques is applicable rnutatis mutandis 
to the 2 d. 

Diffusion from 2 Opposite Parallel Sources.—technique. 
This technique was described by Ouchterlony (22). Except Avith 
regard to its concentration, the agar used in the various techniques 
of double diffusion may be prepared as described for simple diffu¬ 

sion in tubes. Ouchterlony suggests that the agar, at a concentra¬ 
tion of 60 g/ 1 , be clarified by precipitation with calcium chloride, 
filtered through glass wool and, after solidification, cut into small 
cubes and washed with running tap water for 3 days, then with dis¬ 
tilled water, and finally diluted while hot to V 4 its initial concen¬ 
tration. This dilution is made after adding a (iiiantity of NaCl such 
that its final concentration is 0.8 per cent. The final concentiation 
of agar is therefore appreciably less than 1.5 per cent. 

The agar is poured in 2 successive layers into a Petri dish of 10 
cm diameter. One first pours a thin layer which is allowed to so¬ 
lidify Two metallic molds, 15 X 35 or 10 X 10 mm, covered with a 
very thin layer of paraffin, are then placed parallel to each other at 
opposing sides of a rectangle and 20 mm apart on the 1 st layer of 
agar (see Fig. 20 , A). A 2 d layer of agar about 3 mm deep |hen 
pmired around these molds. After the agar has solidified the 2 
^nlrf^ qre verv carefully removed. The reservoirs formed b\ this 
rocedu e are tha! filled with the reactants. “Eve.y 12th or 24th 
h^r depen.ling on how rapidly the <liffusion the . . . . aga 
a^s plL,” tL reservoirs are refilled. Ouchterlony carries out 



the reaction at 37 The plates can be examined in oblictue light 
under a microscope giving 10-fold magnification. Eventual Quan- 
titative measurements are miide by means of a graduated mechani¬ 
cal stage and a cross-hair eye-piece. The plates can be photo¬ 

THEORY .\ND RESULTS.— The Same simplifying hypotheses pre¬ 
sented for simple diflusion are applicable here, with the obvious 







0 ) 







doulile diffusion at given time t. Concentrations 
C and t (/5) of the 2 leactants A and B are expressed in equivalent units 
Crossing of the 2 curves indicates the point of equivalence E. Ce is the concen¬ 
tration of the 2 reactants at this point. Curves C{A) and C(B) were drawn 
following the theoretical laws of diffusion, which is only justifiable as long as Ce 
can be considered negligible w-,th respect to initial concentrations; the dash-line 

portion ot the 2 curves therefore has no precise significance. It is evident that 
Le increases in the course of time. cviucnu uiat 

exception that the concentration of neither reactant remains con- 

as Ouchterlony has pointed out (23 26) 
the diffusion of each reactant can, witiiout significant error be 
considered as diffusion in 1 dimension at least along the line which 
joins the center of each source, neglecting the fact that the depth 
of the reservoir is less than the thickness of the agar layer The con 
cent^of each of the reactants at any instan! and at any pc^nt’ 

were carried out at 

a dish of small diametS7avors the 

of the diffusing molecules by the walls of the vessel reflection 

by deviation at the extremities of 2one.s of precfoitlhli nianifested 

reach nearly to the walls of the dish. Precipitation sufficiently extended as to 


in the gel, where the combination with the other reactant has not 
yet taken place, should theoretically be described by either equa¬ 
tion (2) or eciuation (3), which relate the concentration of the re¬ 
actant (considered constant at the interface between solution and 
gel) and its coefficient of diffusion (Fig. 18). The precijjitate will 
appear at those points and at that time when the 2 following con¬ 
ditions are realized: (1) when the concentrations of the reactants 
reach a threshold level sufficient to allow the formation of a visible 
precipitate, and (2) when the ratio of the concentrations of the 2 
reactants favors the formation of insoluble complexes. 

If the initial concentrations of the 2 reactants are in equivalence 
ratio, the theoretical equations predict that the point where the 2 
reactants will meet at equivalent concentrations (defined here as 
“point of equivalence”) is situated at distances from the sources of 
diffusion which are proportional to the square root of the coeffi¬ 
cients of diffusion of the reactants (23, 2(5), aiul that in the course of 
time the concentrations of reactants will remain in the same ratio 
at the point where they first met. As a result, the zone of precipi¬ 
tation, or at least its maximal density, will remain stationary with 

Ouchterlony, using this method, has determined the ratio of the 
coefficient of diffusion of horse antitoxin to that of diphtheria toxin. 
The ratio found was somewhat less than the value determined by 
free diffusion in liquid media. Ouchterlony attributes this differ¬ 
ence to the influence of the gel. In this particular case (where in¬ 
itial concentrations are in ecpiivalence ratio) the precipitation zone 
appears as a thin immobile sheet of precipitate whose density in¬ 
creases as the 2 reactants continue to reach it.§ 

In most cases, antigen and antibody will not be present in equiva¬ 
lent initial concentrations, aiul the i)oint at which precipitation 
first appears will be displaced in a direction away from the source 
of whichever reactant is present in excess. Furthermore, the point 
of ecpiivalence will not remain stationary but will migrate aAvay 
from the source of the reactant which is in excess. 

If the reactant in excess is the antigen, both edges of the zone ot 
precipitation will usually migrate, since the specific precipitate 
can in most cases be dissolved by the strong concentration of ex¬ 
cess antigen diffusing into the zone. The maximum density of the 

SSelft and & their", 

KX'Sis^toV expocted that high acasitivily can be obtained even 
more readily in tubes. 



zone will increase during its migration toward the antibody source. 

If the reactant in excess is antibody, we would expect the edge 
of the zone (jt precipitation nearest the source ol antibody to le- 
main stationary in a precipitin system. Therefore only the edge 
facing the antigen source will migrate and the zone will continue 
to broaden. This is not true for systems of the flocculation type in 
which the specific precipitate is soluble in excess antibody (Fig. 19). 

It is reasonable to assume that for any given antigen and anti¬ 
body the concentrations required to give just visible maximum 
density of precipitate will always l)e the same under similar con- 

!ig. 19. Photograph illustrating behavior of simple systems using double 
diffusion in agar plates (Ouehterlony’s technique). The 2 square reservoirs 
v\ere placed face to face at a distance of 20 mm. The lower contained the anti¬ 
serum and the upper, the antigen. A, B, C: reaction of ovalbumin, at 3 differ¬ 
ent initial concentrations, with homologous rabliit antiserum (appro.ximately 
1 mg .specihcally precipitable N/cc) at constant initial concentration. A, large 
excess antigen; A, antigen close to equivalence concentration; C, large excess 
anti )ody, D, reaction of purified diphtheria toxoid with excess antitoxin 
(same reactants as in Figs. 10 and 11) (33), 

dition.s." Tlie time retiuired for the concentrations of antigen and 
antibody to attain the tlireshold value recpiired for precipitation at 
a point of the gel will be proportional to the stpiare of the distance 
lietween the sources according to the huvs of diffusion. This rela¬ 
tionship would always be true for the time of appearance of the 
zone of precipitation if the time for formation of precipitate could 
be neglectec . Furthermore, the time for the appearance of pre¬ 
cipitation IS longer when the initial concentrations are weaker 
Ihcorehcal Basis for Determination of Number of Reacting Anti- 

covLn,, whicli showcl each zcne of precipitation 

cone.spoiulb to a smule antigen-antibody reaction in tire case of 

horse aiititoxui system b)°tial'Ldctroi (23," 24 )!™^'°'”* ‘•'■''“'“'''a ‘oai"- 


simple diffusion is not valid in the case of double diffusion. Ouch- 
terlony does not exclude the possibility “that midtiple streaks may 
be formed by a single antigen which reacts at different points in the 
diffusion medium with various antibodies with different diffusion 
rates and reactivity” (22). A priori, this is not likely, for it seems 
logical to suppose that for a single precipitating system the point 
of equivalence divides the gel into 2 regions: that of excess of anti¬ 
gen, where all the molecules of antibody are saturated liy antigen, 
and that of excess ol antibody, where all the molecules of antigen 
are saturated by antibody. It is to be e.xpected that at the point of 
cciuivalence which separates the 2 regions both antigen and anti¬ 
body are completely immobilized by their mutual combination 
and precipitation. For this reason neither can pass into the region 
containing the other.* Only experimental study of a number of 
isolated knowm precipitating systems can resolve this question, as 
was done in the case of simple diffusion. The simple systems studied 
to prepare the illustrations for this chapter have each given a single 
zone of precipitation, with certain exceptions, the causes of error 
for which are kno\\'n and are discussed later (p. 372). 

The independence of zones of precipitation due to distinct anti¬ 
gens which has been established for simple diffusion (p. 353) is ob¬ 
viously not restricted to any 1 type of diffusion and therefore 
should be applicable to the present case. Therefore, the zones due 
to tlistinct antigen in a multiple system should appear, after a 
\'ariable delay, at different points in the medium, as observed by 
()uchterlony, anti each of these zones should remain immobile or 
be displaced toward the sour(*e of antigen or of antibody, depending 
on which is in excess. 

Identification can be carried out by the absorption technitiue 
discussed for simple diffusion. 

Diffusion from 2 Perpendicul.vr Sources.—techniques. 
Two techniques have been described. 

Ouchterlony (22) prepares his agar plates as already described 
(p. 360), either by placing the same molds, 15 X 35 mm, at right 
angles or by using 3 square molds of 10 mm on a side t and mountetl 
together in fixed position (Pig. 20, B and C); one can use all 3 
reseiwoirs (see p. 370) or only 2 of them by filling the 3d with agar. 

Elek (8) chooses as his reservoirs strips of filter paper. For ex¬ 
ample, a strip of filter paper 10 X 60 mm, containing exactly 0.1 
cc of diphtheria antitoxin (4,000 units/cc), is immerseil in the agar 

* Insofar as this reasoning is valid here, it is also valid for simple diffusion, where 
““ by a SaTSto avoid sucking up agar when the 

assembly is withdrawn. 



while it is still liquid. Once the agar has solidified a similar strip 
containing 0.1 cc of concentrated antigen solution (e.g., diphtheria 
toxin, 4,000 Lf units/cc) is placed on the surface of the gel, forming 
a right angle with the preceding strip. After a convenient time, the 
plates may be placed on photographic paper and contact prints 

THEORETICAL ANALYSIS.! To Understand what takes place in the 
gel as a whole, let us assume that at a given distance the concen¬ 
tration of each reactant is equal at all points lying along a line 
parallel to its source of diffusion. This assumption would only he 
fully justified if the source were long and if one were to neglect 
points lying near its extremities. It is considered sufficient, how¬ 
ever, to describe the distribution of concentrations of the reactant 
along a single perpendicular to its source to understand what 

for 1, 2, 3 and 4, 15 X 35 min; for 5, 6 and 7, 10 X 10 mm; distance from .4 
to B and from E to F, 20 mm. Different variations of form, of number and of 
arrangement of reservoirs can naturally he imagined (after Ouchterlony (22)). 

happens along other perpendiculars. Furthermore, the diffusion 
along each perpendicular may be considered as simple diffusion. 

Let us finally assume that the concentration of each reactant 
remains constant in the neighborhood of its source so that we ma^^ 
apply either equation (2) or (3). In practice, this condition is far 
from true. 

The curves shown in Figure 21 were constructed after making 
the above simplifying assumptions and neglecting all effects due 
to mutual combination and precipitation of the reactants. The 

sources of diffusion are drawn contiguous at their extremities in 
Figure 21. 

Under the ideal conditions so defined, the specific precipitation 
of the reactants during the course of their diffusion vdll resemble 
to a certain extent what we have seen in simple diffusion and will 
resemble even more closely double diffusion from 2 parallel sources. 

soS.VdXSSnerlsK" Elek in a 

Concentrations (B) 

Fig. 21 (above ).—Schemutic ropresentutioii of double diffusion Iroin 2 per¬ 
pendicular sources at a given time t. Consider B as antibody, and .li and Ao 
as 2 antigens with distinct specificity but the same diffusion coefficient. One 
assumes that antibodies anti-/li and anti-A 2 are at the same initial concentra¬ 
tion and have the same coefficient of diffusion, which is lower than that of Ai 
and A 2 ; their concentrations are therefore represented by the same curve 

C’urves of the concentrations conform to the theoretical laws of diffusion in 
1 dimension, with the hypothesis that <liffusion of each reactant along each 
perpendicular to its source conforms to thpe laws. Because of the conventional 
choice of equivalent units of concentration of antigen and antibody (as for 
graphs in Figs. 1 and 18), each line of equivalence descrilK^s the locus of points 
where concentrations of antigen and antibody are equal. To give an idea of the 
variable density of precijiitate, these loci are represented by tracings whose 




Let us consider the diffusion of 1 reactant, A, along any perpen¬ 
dicular to its source. We may assume that at all points along such a 
perpendicular the concentration of the other reactant, B, will be 
equal at any given time t but will increase with time. If we now con¬ 
sider perpendiculars brought closer and closer to the source of re¬ 
actant B, then at the limit the concentration of reactant B not only 
Avill be equal along the perpendicular at time t but will remain con¬ 
stant with time; these are precisely the theoretical conditions en¬ 
countered in simple diffusion, and the conclusions which follow 
from the theoretical study of simple diffusion should apply to this 
limiting case as well. In particular, the point of equivalence in the 
gel is very close to the intersection of the 2 sources at time / = 0 + 
e no matter what the initial ratio of antigen and antibody concen¬ 


Aside from the limiting case, one may apply what has been said 
regarding double diffusion from 2 parallel sources to the condition 
in which the sources are perpendicular by simply considering for 
each point in the gel its perpendicular distances to the sources of 

When both reaetants have equivalent initial concentrations (Fig. 21; 
reactants A 2 and B) the perpendicular distances from the points 
of equivalence to each of the 2 sources of diffusion will remain 
in constant ratio, the value of which will be the same as that 
found for double diffusion from 2 parallel sources. In this particular 
instance, the geometric locus of the equivalence points will be a 
segment of a straight line originating at the intersection of the 2 
sources of diffusion. The angles formed between this line and the 
sources of diffusion will remain constant with time. The ratio of the 
sines of these angles will have the same value as would the ratio of 
the distances from the point of equivalence to the sources of dif¬ 
fusion in the case in which the same reactants are diffusing from 

thickness is roughly proportional to concentrations of the reactants at each 

The initial concentrations of /U and anti-Ao (B) are in the ratio of equiva¬ 
lence (equal on the graph). The line of equivalence is straight and passes 
through the intersection of the sources. This straight line does not bisect the 
angle between the sources, because the coefficients of diffusion are different for 

/I 2 tinci cinti~2i2 (,*^/« 

A t '"le of equivalence of Ai (different from those of 

iespec-^o L b of A.) with 

^ anti-Az {B) 1 of the 2 reactants being in e.xcess, the line of equiva- 

FT/'29^r/ intersection of the sources only at the time 0 + e (33) 

b iG. 22 (Wow;).—Conditions are the same as in Figure 21, with 1 e.xcentioir 
l ientical antigens, instead of 2 different ones, diffuse from sources A. and U 

herSr.r'“‘"“ 21)- of the points of equitdence is 

centritions concentration of antigen the sum of the con- 

«« ->-ence of 


parallel sources ;§ in other words, the same value as the ratio of the 
square roots of the diffusion coefficients. 

The maximum of the density of precipitate for each perpendicu¬ 
lar from I or the other source of diffusion will be at a point close to 
the point of equivalence. This maximum density will increase with 
time at any given point, and at any given time will become less and 
less as one considers points in the gel farther and farther from the 
sources of diffusion until finally no precipitate is visible. In analogy 
to what we have seen for diffusion from 2 parallel sources, the point 
at which the zone of precipitate is no longer visible will be dis¬ 
placed in such a manner that the perpendicular distances from this 
point to the sources of diffusion would be proportional to the square 
root of time if the time of formation of the precipitate coidd be 

When the reactants are not present in equivalent concentrations 
(Fig. 21; reactants Ai in excess and B) the zone of precipita¬ 
tion will be displaced in a direction away from the source 
of diffusion of the reactant that is present in excess. The appearance 
of the zone of precipitation will differ from that of the preceding 
case as woidd be expected from what has been said regarding dif¬ 
fusion from 2 parallel sources (see p. 362 and Fig. 19). Naturally, 
here as in the preceding case the density of precipitate will diminish 
at points farther and farther away from the sources of diffusion. 

PRACTICAL CONSIDERATIONS. —In practice, experimental condi¬ 
tions for double diffusion at right angles deviate from the ideal ones 
discussed above according to the particular technique chosen. The 
most important differences are: (1) The 2 sources cannot be ex¬ 
actly contiguous as in Figure 21. (2) The concentration of the re¬ 
actants will not remain constant at the sources of diffusion if the 
volume of the reservoirs into which they are placed is small in com¬ 
parison with the volume of gel into which they are diffusing." 
(3) Of necessity, the depth of the reservoirs must be less than the 
thickness of the agar. 

Ouchterlony’s technique deviates from the ideal particularly vith 
respect to points (1) and (2). The deviation is less serious vitli re¬ 
spect to point (3) since the lower layer of agar is usually thinnei 
than the upper layer. Because of the appreciable distance that anti¬ 
gen and antibody must diffuse (even from the points closest to¬ 
gether between the 2 sources of diffusion), an appreciable time will 

§ This reasoning finds experimental s'^'tS^a^^es 

lonv (22): the distances found are, respectively, 8.5 and -0 8.5 . . ^ 

.87° and 90° - 37° = 53°; the ratio of the distances is thus 0.74 and ot the sines 

ll^buchterlony (23) notes tha,t the concentration of reactants in his reservoirs de- 
craases about 25 per cent within 12 hr. 



elapse before concentrations are built up sufficiently high to result 
in visible precipitation. If the initial concentrations of the reac¬ 
tants are low, 1 or more days may be necessary before the appear¬ 
ance of a precipitate. In a multiple system this period will, of 
course, be different for each of the different zones of precipitation. 
Since the reseiwoirs do not touch each other, the zone of precipita¬ 
tion will first appear and remain most dense in the space which 
separates them and will form an obtuse angle since we must now 
consider diffusion from adjacent sides of the same reservoirs. 

The periodic refilling of the reservoirs required by Ouchterlony’s 
original technique compensates to some extent for the fall in con¬ 
centration of the reactants within the reservoirs. One modification 
suggested by Ouchterlony eliminates the necessity for refilling the 
reservoirs by sealing the Petri dishes with Plasticene to prevent 
evaporation. Although this modification leads to somewhat differ¬ 
ent results and deviates from ideal conditions, it does not appear 
to considerably the value of the qualitative findings. For 
reasons to be discussed, it seems to us preferable to avoid refillings 
in qualitative experiments. 

Elek’s technique deviates from ideal conditions particularly with 
regarti to points (2) and (3). The reservoirs are particularly small 
in this case since their content is “exactly 0.1 ml.” for a filter paper 
strip 10 X 60 mm. Presumably it is for this reason that Elek uses 
antitoxin and toxoid of very high titer (4,000 units/cc). For the 
same reason, in his analyses of human serum Elek found it neces¬ 
sary to dry successive quantities of the immune serum in 3 strips 
of filter paper which were then superimposed to serve as the reser¬ 
voir of antibody.* 

The very rapid fall in concentration in reservoirs of such small 
volume probably explains also the stationary “planes of floccula¬ 
tion” described by Elek, an observation quite contrary to the find¬ 
ings 1 ) 3 ^ Ouchteilony s te(‘hnique, particularly when the antigens are 
as highly concentrated as certain of the constituents present in 
seiinn. In Elek s procedure the depth of the reservoirs containing 
the reactants is very small compared with the thickness of the agar 
Moreover, the filter paper strips are not held in the same plane so 
that they are separated by a thin layer of agar near the vertex of 
the angle which they form. The geometric locus of the points of 
equivalence in this region will be a plane parallel to the strips and 

day with fresh immune senini "S ivoro rvK+r.' 5 could be moistened each 
less than that obtained by other methods (( 28 ^ 31 *^^ V^^”i^ deal 
dure could be still further perfected W). Perhaps this proce- 

volume o, reactmft' “ * 


\vill, of course, pass hetAveen them. In regions at a distance from 
t le sources ot diffusion this locus will become a plane perpendicular 
to the gel; it thus undergoes a rotation and twisting in the neighbor¬ 
hood of the angle formed by the sources (8). 

Diffusion of Same Rcuctont from 2 Parallel Sources,' Coalescence 
of Zones. Diffusion in 2 dimensions allows one to place 2 or more 

Fig. 23 (above ).—Photograph of zones of precipitation of 2 diffusing anti¬ 
gens (Ouchterlony’s te(dinique) under conditions as as po.ssihle to the 
theoretical conditions of Figure 21. The 2 antigens were ovalbumin (At) 
whose initial concentration is in equivalence with that of the antibody, and 
human serum albumin (.b) at 3 times the concentration of the ovalbumin. The 
immune serum (B) is a mixture such that the same volume jirecipitates at 
equivalence with equal quantities (in terms of protein N) oi the 2 albumins 

Fig. 24 (below ).—Same conditions as for Figure 23 except that Ai contain.s 
ovalbumin at an initial concentration 3 times its concentration in Ao and 
equal to that of human serum albumin used in Figure 23. Note coalescence of 
zones of precipitation. The experimental conditions apj)roximate as close as 
possible to the theoretical conditions of Figure 22 (33). 

sources of diffusing reactants in the same gel (22, 8). ^\ e A\ill limit 
ourselves to the simplest case, that in which 2 sources of reactants 
are placed parallel to each other and perpendicular to a 3d. For 
example, 2 different solutions containing antigens m common 
could be placed at the parallel sources and antiserum at the per- 

pendietdar source. . , * i 

The graphs shown in Figures 21 and 22 describe the expected 

behavior using 2 distinct antigens (Fig. 21) and identical anbgen.s 
(Fig. 22) at the 2 parallel sources. If the 2 antigens are of entire } 



different specificity, each Avill react with its respective antibody in¬ 
dependently of the othert (p. 353 and Fig. 14). The geometric loci 
of their respective points of equivalence will cross without disturb¬ 
ing each other, as ob.served by Ouchterlony and Elek (Figs. 21 and 

If, under similar conditions, the same antigen is placed at both 
parallel sources, then at each point in the gel the concentration of 
this antigen will be the sum of the amounts which have diffused 
from each source. The resulting loci of the points of equivalence 
will therefore be displaced and bent toward the opposing source. 
The displacement of the locus from 1 source will increase as the other 
source is approached, anrl the effect of the concentration of antigen 
diffusing from the opposite source becomes more and more pro¬ 
nounced. Instead of crossing, the loci of the points of equivalence 
Avill eventually join and form an arc| (Figs. 22 and 24). 

As Ouchterlony has pointed out: (1) if the concentration of a 
particular antigen is the same at both sources, the coalescing zones 
will be arranged symmetrically, and (2) an asymmetrical arrange¬ 
ment of the zones indicates that the highest concentration of anti¬ 
gen is present in the reservoir situated farthest from the point of 

Partial coalescence: When antigens which cross-react are placed 
at the parallel sources of diffusion and an antibody homologous to 
1 but not to the other is placed at the perpendicular source, Ouch¬ 
terlony has observed in the author’s laboratory a partial coales¬ 
cence of the type shown in Figure 25. 

APPLic'ATioN TO iMMiTNocHEMic.\L ANALYSIS.— Dclertninatiou of 
.Minimum Number of A nligens in a System; Sensitivity of rechniques 
of Double Dflfusion—\Ye have considered the theoretical basis 
foi determination of the number of reacting antigens by means of 
( ouble diffusion (p. 3C3). In practice, the ease of enumeration will 
depend on whether it is of the flocculating type, which will usually 
give sharp zones when the reactant in excess is antibody or 
whether it is of the precipitin type, which will give sharp zones 
only when the initial concentrations of antigen and antibody are 
close to their equivalence ratio. It should be recalled that at low 
initial concentrations ot antigen and antibody the appearance of 
e zone of precipitation will be delayed and that the zone given by 
anjmt^whose initial concentration is in great excess over that 

as evidence eithe^'^for^oragaiLt'^tlie UieorVlhar^r**^'.:!*^' cannot be used 

specific, as Elek has claimed (8). ' ^ of precipitation is 

Elek of ‘looping” and otfiers of ‘‘fusion ” It ic speaks of interference,” 

to the phenomenon observed by Petrie. ^ clear that all of these terms pertain 


of antibody may reach the antii)ody source within a few days. For 
this reason the plate should be examined at intervals so that no 
zone escapes detection. 

Under ideal conditions lor diffusion from perpendicular sources 
(when sources are contiguous and concentrations of reactants re¬ 
main constant at the sources), one might expect the sensitivity to 
be of the same order of magnitude as that pertaining to simple dif¬ 
fusion (at least in the regions closest to the sources, see p. 367). 
In practice, the exact limit is hard to define because so many vari¬ 
ables must be considered. Thus in Ouchterlony’s technique the 
sources of diffusion are not contiguous, time plays an important 
role and the densities of the zones which are faint at 1 st later in- 

Fig. 25.—Cross-reaction of 2 antigens with the same serum: chicken ovalbumin 
(A 2 ), duck ovalbumin (Ai) and anti-chicken ovalbumin rabbit serum (B). 

crease in a manner which depends on the respective concentrations 
of the antigen and antibody. Moreover, since the rate at which the 
concentration of the reactants diminishes depends on their volume, 
the limit of sensitivity in double diffusion will depend on the size 
of the reservoirs. As we have already seen, this applies especially 

to Elek’s technique. 

Sources of error: While carrying out reactions according to Ouch¬ 
terlony’s technique to obtain illustrations for this chapter, the 
author observed immobile false boundaries on several occasions. 
It is important to avoid anomalies of this sort since immobility is 
not a criterion for differentiation between false and true bound¬ 
aries, as in the case of simple diffusion. The false boundaries 
obseiwed were presumably caused by repeated refilling of the resei- 
voirs, since they have not been observed in sealed Petri dishes. 
Formation of false boundaries is particularly likely if the 1st layer 
of agar fails to cover the curved portion of the bottom of the I etri 
dish sufficiently. However, they have been observed even when the 
surface of the 1 st layer of agar was perfectly plane (Fig 26). 

In general, it would appear wise to avoid any procedure wine 1 



might cause fluctuations in concentrations of reactants in the reser¬ 
voirs such as refilling, or abrupt changes in temperature, although 
as yet, specific information is lacking with regard to the latter 

Identification .—The coalescence of zones provides a means for 
identification of antigens.§ If 2 antigens diffusing from parallel 
sources form zones of precipitate which coalesce completely, they 
may be considered identical; if the zones cross, the 2 antigens are 
different. Naturally, as long as the zones of precipitation fail to 
reach one another, we can conclude nothing. This may occur in the 
particular instance in which antibody concentration is very low 

tiG. 26. Artefacts due to repeated refillings of reservoirs. Reactants: oval¬ 
bumin at 2 different initial concentrations (Ai and A 2 ) and homologous rabbit 
serum (fi). Reservoirs were refilled at 1, 2, 3, 4 and 6 davs after start of the ex¬ 
periment. Photograph taken on 7th day (33). 

and the concentrations of the antigens are very high, so that the 2 
zones of precipitation are short, move quickly away from the 
sources of antigen and reach the antibody source before their ex¬ 
tremities have united. Furthermore, Elek states that “lack of 
looping and consequently erroneous interpretation—may occur if 
°already well formed when a second approaches 
It (8). Aside from this infrequent condition, the plates are usually 
easy to read it the number of zones of precijiitation is not too great 
especially when the antigens diffusing from the 2 sources are identi- 


Identification can also be made using absorbed senim as Ouch- 
terlony has done with diphtheria toxin (22). In I liis case it shotdd be 
remembered that the density of the zones and the relative distances 
betw een them and the sources will vary with time. It is well to com- 

dever®®T ‘‘‘f (or their photographs) before and after absorption 
developed under identical conditions and for the same length of 

sourS;”' reactant which diffuses from the perpendicular 



HeiG we shall discuss the behavior of antigens diffusing into the 
gel fiom bacterial colonies growing upon it. Obviously the tem¬ 
perature, composition of the medium, etc., must be suitable tor 
production of the antigens under investigation. Such techniques 
have been used from time to time since they w’ere described by Sia 
and Chung (37) and Petrie (34) in 1932. The techniques may be 
considered either as simple diffusion or double diffusion in 2 dimen¬ 

Simple diffusion: Immune serum is incorporated in appropriate 
concentration with the agar medium which is then poured into 
Petri dishes." For methods of inoculation of the plates the reader 
is referred to the original articles (37, 34, 13, 35, 19-21). 

It follow's from the studies of simple diffusion that the zones of 
precipitation which may appear around a given bacterial colony 
will be less and less dense and more and more distant from the 
colony as the antibody content in the medium is decreased. 

Double diffusion: Two similar techniques using agar in Petri 
dishes have been described by Ouchterlony (19-21, 25) and by 
Elek (7-9). Ouchterlony cuts out a transverse trench of agar and 
refills it wdth diphtheria antitoxin or antidiphtheria serum incor¬ 
porated in agar. He then streaks 1 or more different strains of the 
diphtheria bacillus perpendicular to the trench. Elek allows a 
sterile strip of filter paper soaked in antitoxin to sink into the me¬ 
dium w’hile the agar is still licpiid. After the agar has hardened the 
organisms are streaked in parallel lines across the plate and per¬ 
pendicular to the filter paper. 

In both techniques zones of precipitation develop in a manner 
analogous to those discussed under double diffusion from perpen¬ 
dicular sources. The zones of precipitation commence at or near the 
point w'here the streaks of culture join the antitoxin source and 
form an acute angle wdth the streaks. If 2 zones of precipitation 
coming from neighboring streaks coalesce, it may be concluded that 
they are due to identical antigensj if the zones cross, the antigens 
must be different. 

Various means may be used to identify w hich zone is due to pre¬ 
cipitation of toxin itself. These include the time at which the zone 
appears, c(mlescence with a known zone from a known toxi^mc 
strain or from a reservoir containing toxin, and dilution of the 
antitoxin so that zones caused by precipitation of nontoxic antigens 
will disappear. 

II This type of experiment has not been carried out in tubes. Perhaps it may prove 
of value in the case of anaerobic organisms. 



Elek has applied the same technique to staphylococcus and 
Clostridium welchii. 

All of these techniques, whether concerned with simi)le or with 
double diffusion, differ in 1 important respect from the case in 
which the antigens diffuse into the gel from a solution. In the latter, 
the antigen concentration will diminish continuously in the reser¬ 
voir unless it is replenished. When antigen diffuses from a bacterial 
colony, the rate at which it is released is not known, regardless of 
whether its release is due to actual excretion of antigen or to bac¬ 
terial autolysis. If one wishes to determine the number of antigens 
released, it must be kept in mind that variations in the concentra¬ 
tion of antigen diffusing from the source may give rise to errors. 
Other sources of error must also be considered, particularly the 
possibility of nonspecific precipitate with constituents of serum, 
a possibility that necessitates the use of normal serum controls. 


With each diffusion technicpie we have described general meth¬ 
ods for identification of the individual zones of precipitation given 
by a mixture of antigens (see pages 35(5 and 373). It is also possible 
that information provided by determination of the diffusion co¬ 
efficients (pp. 350 and 3(52) may be used to assist in identification. 
Finally, certain antigens endowed with enzymatic activity may be 
identified by means of visil)le reactions with a substrate incorpo¬ 
rated in the gel (8). For example, Elek has cultivated CL welchii on 
agar plates containing human serum. When commercial serum 
was allowed to diffuse through the gel, Elek observed that the limit 
of opacity caused by a-toxin diffusing from the colonies (Nagler re¬ 
action) (;oincided with I of 2 zones of precipitation. He therefore 
identified this zone as being caused by the a-toxin. Elek also inves¬ 
tigated staphylococcal hemolysin in an analogous manner. 

It would seem, a priori, that any of the techniques of simple or 
double diffusion might be applied to the study of solutions of anti- 
pns with enzymatic activity as well as to those released from living 
bacterial cultures. In many cases it is not even necessary to incor¬ 
porate the substrates in the gel until after the diffusion pattern 
has developed. 

Cohn (5) has carried out an experiment of this tvpe using the B- 
galactosidase of Escherichia coli as antigen with its homologous 
labbit antiserum. The double diffusion technifiue of Ouchter- 
lony was fo lowed After (5 days, the substrate, o-nitrophenyl B- 
gci actosK e (NPCt) was sprayed over the agar surface. An intense 
ye\\o^^ color was produced by hydrolysis of NPG in the region of 
antigen excess which was limited by the zone of precipitation. 


This method of identification can be further extended for general 
use when the product of enzyme action, though not itself visible, 
can be rendered visible at a convenient time by addition of a suit¬ 
able indicator. 

In using diffusion methods for identifying enzyme-antienzyme 
systems, many factors will require consideration. These will vary 
with each system investigated and will depend on: the diffusibility 
ot the leaction products, whether or not enzyme activity is in¬ 
hibited by specific antibody, the time when the substrate and per¬ 
haps indicator are added to the medium, whether the initial con¬ 
centration of antigenic enzyme is in excess relative to antienzyme 
or vice versa, etc., etc. 

It should be stressed that if antibody (antienzyme) is in excess, 
the limit of enzyme-substrate reaction may not necessarily coincide 
with the antigen-antibody equivalence point and consequently 
may not coincide with the zone of precipitation in a flocculation- 
type reaction. 


1 . Bechhold, H.; Structurbildung in Gallerten, Ztschr. phys. Chemie 52: 

185-199, 1905. 

2. Becker, E. L.; Munoz, J.; Lapresle, C., and Lebeau, L.: Antigen-antibody 

reactions in agar: II. Diffusion coefficients of antigens, Federation 
Proc. 10: 401, 1951. 

3. Becker, E. L.; Munoz, J.; Lapresle, C., and Lebeau, L.: Antigen-antibody 

reactions in agar: 11. Elementary theory and determination of diffusion 
coefficients of antigen, J. Immunol. 67: 501-511, 1951. 

4. Brown, R.: Rhythmic precipitation of pneumococcal soluble specific sub¬ 

stances and antipneumococcal sera, Proc. Soc. Exper. Biol. & Med. 45: 
93-95, October, 1940. 

5 . Cohn, M.: Unpublished results. 

6 . Duclaux, J.: TraiU de Chimie Physique appliqu^e a la Biologie (Paris: 

Hermann et Cie, 1938), Vol. 11. 

7 . Elek, S. D.: The recognition of toxicogenic bacterial strains in vitro, 

Brit. M. J. 1: 493-496, Mar. 13, 1948. 

8 . Elek, S. D.: The serological analysis of mixed flocculating systems by 

means of diffusion gradients, Brit. J. Exper. Path. 30: 484-500, Octo¬ 
ber, 1949. 

9 . Elek, S. D.: The plate virulence test for diphtheria, J. Clin. Path. 2: 250- 


10. Hanks, J. H.: A ring precipitation test for estimating the concentration 

of antibody in small amounts of immune serum, J. Immunol. 28. 95- 

104, 1935. „ , 

11 . Hedges, E. S.: Liesegang Rings and Other Periodic Structures (London. 

Chapman & Hall, Ltd., 1932). „ . /r- 

12. Hitchcock, D. 1. in: Hober, R., et al.: Physical Chemistry of Cells and 1 is¬ 

sues (Philadelphia: Blakiston Company, 1946), Section 1. 

13. Kirkbride, M. B., and Cohen, S. M.: Precipitation reactions of mening^ 

coccus strains with immune serum in agar plates in relation to antigenic 
activity. Am. J. Hyg. 20: 444-453, 1934. 








20 . 
21 . 

22 . 



20 . 










Maegraith, B. G.: Rough and smooth variants in stock cultures of menin¬ 
gococci, Brit. J. Ex])er. Path. 14: 227—235, 1933. 

Mayer, M., and Ileidelljerger, M.: Velocity of combination of antibody 
with specific polysaccharides of pneumococcus, J. Biol. Chem. 143: 
567-574, 1942. 

Mitchison, D. A., and Spicer, C. C.: A method for estimating streptomy¬ 
cin in serum and other body fluids by diffusion through agar enclosed 
in glass tubes, J. Gen. Microbiol. 3: 184-203, 1949. 

Munoz, J., and Becker, E. L.: Antigen-antibody reactions in agar: I. 
C'omplexity of antigen-antibody systems as demonstrated by a serum- 
agar technic, J. Immunol. 65: 47-58, 19.50. 

Nicolle, M.; Cesari, E., and Debains, 10.: Etudes sur la iirecipitation des 
anticorps et des antigenes (deuxieme mcmoire): Serums antitoxiipies, 
Ann. Inst. I>steur 34: 596-599, 1920. 

Ouchterlony, O.: In vitro method for ti'sting t he toxin-producing cajiacity 
of diphtheria bacteria, .\cta jiat h. et inicrobiol. scandinav. 25: 186-191, 


Ouchterlony, O.: Antigen-antilxxly reactions in gels, Acta path, et micro- 
biol. scandinav. 26: 507-515, 1949. 

Ouchterlony, O.: In vitro method for testing the toxin-jiroducing capacity 
of diphtheria bacteria, Acta path, et inicrobiol. scandinav. 26: 516-524, 

Ouchterlony, O.: Antigen-antibody reactions in gels, Arkiv Kemi, Mineral. 
Geol., Vol. B26, no. 14, pp. 1-9, 1949. 

Ouchterlony, O.: Antigen-antibody reactions in gels: II. Factors deter¬ 
mining the site of the precipitate, Arkiv Kemi, Mineral. Geol., Vol. 1, 
No. 7, pp. 43-48, 1950. 

Ouchterlony, O.: Antigen-antibodj' reactions in gels: III. The time factor, 
Arkiv Kemi,Mineral. Geol., No. 9, pp. 55-59, 1950. 

Ouchterlony, O.: An in vitro test of the toxin-producing capacity of 
Corynebacterium diphtheriae, Lancet 2: 346-348, 1949. 

Ouchterlony, O.: “Antigen-antibody Reactions in Gels and the Practical 
Application of This Phenomenon in the Laboratory Diagnosis of 
Diphtheria.” Doctoral Thesis in Medicine, Stockholm, 1949. 

Oudin, J.: Methode d’analyse immunochimique par precipitation spe- 
cifique en milieu gelifie, Compt. rend. Acad. sc. 222: 115-116, 1946. 

Oudin, J.: L analyse immunochimique du s^rum de cheval par precipita¬ 
tion specifique en milieu gelifie: Premiers resultats, Proc. VHth Cong, 
de chim. biol. Li6ge, Oct. 3, 1946, pp. 10-11; Bull. Soc. chim. biol. 29: 
140-149, 1947. 

Oudin, J.: L’analyse immunochimique qualitative: Methode par diffusion 
des antigenes au sein de rimmuns6rum iirecipitant gelose; premiere 
partie, Ann. Inst. Pasteur 75: 30-52, 1948. 

Oudin, J.: L’analyse iriimunochimique qualitative; IMethode par tliffusion 
des antigenes au sein de rimmunserum precipitant gelos6; Deuxieme 
partie, Ann. Inst. Pasteur 75: 109-130, 1948. 

Oudin, J. : La diffusion d’un antigene dans une colonne de gel contenant 
les anticorps precipitants homologues; Etude quantitative ties trois 
prmcipales variables, Compt. rend. Acad. sc. 228: 1890-1892, 1949 

Oudin J.: “L’analyse immunochimique: Expose critique d’une iiK^thode- 
application au s6rum de cheval et au lait de jument.” ThKse de Doctorat 
es Sciences, Pans, 1949. 

Oudin, J.: Unpublished work. 

Petrie, G. F.: A specific precipitin reaction associated with the growth on 





agar plates of meningococcus, pneumococcus and R. dysenteriae (Shiga) 
Bnt. J. Exper. Path. 13; 380-394, 1932. ’ 

Petrie, G F., and Steahben, 1).: Specific identification of the chief i)atho- 
gemc Clostridia of gas gangrene, Brit. M. J. 1; 377-379, 1943. 

Pope, C. G. ; Stevens, M. F.; Caspary, E. A., and Fenton, E. L.: Some new 

observations on diphtheria toxin and antitoxin, Brit J Exner Path 
32:246-258,1951. ■ y . ciiu. 

Reinei, L., und Kopp, H.: Ueber Entstehung von Liesegang’schen Rin- 
gen bei der serologischen Prazipitation, Kolloid Ztschr 42- 335-338 
1927. ’ 

3/. Sia, R. H., and Chung, S. I.: ITse of antipncumococcus-serum-agar for 
the identification of pneumococcal types, Proc. Soc Exiier Biol & 
Med. 29: 792-795, 1932. 

38. Taboury, M. I., and daboury, I. J.: Li' {ihenomene periodique de Liese- 

gang et la structure des jiroteines, J. chim. phys. 41: 89-99, 1944. 

39. Veil, S.: Les phinomenes periodiques de la chimie: I. Les periodicites de 

structure (Paris: Hermann et Cie, 1934). 


[Page numbers printed in bold face indicate original contributions to this volume.] 


Al)s<)ri)tion: in coniiwund identifica¬ 
tion in simiiie prociititating 
systems in }j;e]s, 358 f. 

Acids, amino, see Amino acids 

Acids, organic: analysis 

by countercurrent distrilmtion, 17 

l)y jjaper chromatograithy, 37, 44 

Adsorjttion technitiues, see Chroimi- 
tograj)hy, papei- 

.\gar: prei)aration of for simple dif¬ 
fusion in jjreciititin reaction in 
gels, 340 f. 

.\lkaloids: analysis by paper chroma¬ 
tography, 37 f. 

Amino acids 

analysis by countercurrent distri¬ 
bution, 19 f. 

analysis by jtaper chromatograjihy 
application of, 30 f. 
jjaper for, 35 
(]Uantitative, 49 f., 51 f. 
solvents for, 35 ff. 
detection by paper chromatog¬ 
raphy, 41 f. 

identification by paper chroma¬ 
tography, 45 ff. 
partition ratios, 20 

.\mj)hoteric substances: analysts by 
countercurrent distribution. 19 
ff., 20, 21 

Animal sera, see Sera, animal; 
Lthanol fractionation 

Antibiotics and pai)er chromatog¬ 
raphy, 44 


flocculating, in simi)le precijjitation 
reactions in gels, 351 f. 
production in animals, 271 ff. 
animal choice, 271 f., 274 f! 

—injections, 277 ff. 

—preparation, 275 ff. 

—purity, 273 
—serum collection, 279 ff. 

—stability, 273 f. 
test bleedings, 279 

.\ntib()dy-active proteins, see Pro¬ 
teins, antibody-active; Etha¬ 
nol fractionation 
Antidiuretic substances 

in blood—bioassay of, 209 ff. 

procedure, 210 ff. 
in urine 

bioassay of, 201 ff. 

—with dogs, 202 f. 

—with rats, 201 f. 
concentration of, 204 f. 

—procedure for determining, 205 
Antigen, in antigeneity, see Anti- 

Antigen-antibodv reaction, analvsis 
of, 308 ff.‘ 

Antigeneity, factors in, 271 ff. 
animal, 271 f. 
antigen, 273 ff. 

experimental animal choice, 274 


injections, 277 ff. 
preparation, 275 ff. 
purity, 273 

serum collection, 279 ff. 
stability, 273 f. 
test bleedings, 279 

Anti-kidney serum in induction of 
nei)hrotic syndrome, 264 ff. 
Antipyrine: in total body water 
measurement, 170 f. 

Antiserum, rabbit ovalbumin: reac¬ 
tion with (‘hicken ovalbumin 
340 ff. 


analytical ultracentrifuge, 110 ff. 
ascending paper chromatography 
33 ff. H .), 

countercurrent distribution, 7 ff. 
electrophoresis, 04 ff. 
micro-electrophoresis, 101 , 102 
micro|)honic manometer, 251 ff, 
paper chromatography, 33 ff. 
lueparative ultracentrifuge, 120 ff. 
weight-pattern determination, 10 
Arginine Tm: mierobiologic esti- 
matioii(indog), 220 ff. 
argiinne determination, 222, 223 




Arginine Tm (cont.) 

creatinine determinations, 222, 223 
protocol for, 221 

renal-clearance determination, 221 


Assa 3 '^, biologic, see Bioassay 


and quantitative precipitin reac¬ 
tions in gels, 374 f. 
and specific precipitation in gels, 
335 f. 

Bases, organic: analysis by counter¬ 
current distribution, B) 

Bioautography, 44 

Blood: bioassay' of antidiuretic sub¬ 
stances in, 209 ff. 

Blood pressure, arterial: measure¬ 
ment (in rats), 251 ff. 
indirect recording, 253 ff. 
microphonic manometer for meas¬ 
urement of systolic pressure 
(in rat), 251 ff. 

Body water, total: study of 
antipyrine method, 170 f. 
isotope techniques, 168 ff. 

Body water compartments: study' by 
space techniques, 159 ff. 
extracellular fluid, 159 ff. 

analy'tical procedures, 163 ff. 
calibrated infusion technique, 
161 ff., 165 f. 

constant infusion-urine collec¬ 
tion technique, 1601., 165 
intracellular water, 171 ff. 
total body water 

antipyrine method, 170 f. 
isotope techniques, 168 fl. 

Bofulinus toxin, 325 

B phage preparations of E. coli: 
analysis of, 318 

Braun-Menendez and Chiodi pro¬ 
cedure for determination of 
renal function in rats, 248 


for electrophoresis, 87 ff. 
in ethanol fractionation of j)roteins, 
287 f. 


Calculation _ 

in countercurrent distribution, 13 


of diffusion in gels, 337 f. 
in electrolyte excretion studies (in 
dog), 178 . 

of endogenous creatinine clearance, 
216 1- . • Qo f 

of motility in electrophoresis, 82 l. 

for sedimentation after ultracentrif¬ 
ugation, 119 ff. 

Calibrated infusion technique in 
study of extracellular fluid in 
body water compartments, 161 

Catheterization, renal, 147 ff. 


choice for micro-electrophoresis, 
102 f. 

Sliedlovsk.v conductivity', 91 
Tiselius, in electroly te excretion 
study, 176 

(,’entrifugal fields: nature of sedi¬ 
mentation in,107 ff. 

Champy fluid in tissue fixation, 237 
Chromatography', paper, 25-62 
apparatus for, 33 ff. 
applications of, 29 ff, 

apparatus for, 33 ff. 
compound-detection in, 41 ff., 
48 f. 

identification in, 44 ff., 49 
papers for, 34 f. 

preparation and manipulation in, 
39 f., 47 ff. 
solvents for, 35 ff. 
supplies for, 33 ff. 
compound-detection in, 41 ff. 
distinguishing featuies of, 25 
identification in, 44 ff. 
limitations of, 53 f. 

ascending, 33 ff. 
choice of, 32 
quantitative, 49 ff. 
reverse phase, 52 f. 
papers for, 34 f. 

preparation and manipulation in, 
39 f. 

principles of, 25 ff., 27 
ciuantitative, 49 ff. 
reverse phase, 52 f. 
solvents for, 35 ff. 
supplies for, 33 ff. 

Conalbumin, 314, 325 
Conductivity: measurement m elec¬ 
trophoresis, 90 ff. 

Constant infusion-urine collection 
technique, 160 t. 
apparatus. Off. 

Convection: control in electropho¬ 
resis, 93 ff. 

Countercurrent distribution, 3—24 
alternate withdrawal, 12 
calculation of theoretical curves, 
13 ff. 

completion of s(iuares, 12, 15 
distribution series, 3 ff. 
fractionation (simple), 3 fl. 


Countercurrent distril)ution (cord.) 
fuiuhimental procedure, 11, 13 IT. 
recycling, 12 1’., 15 
single withdrawal, 11 f., 15 
systems for 

amino acids, 19 f. 
amphoteric substances, 19 ff. 
definition of, 5 
organic acids, 17 ff. 
organic bases, 19 
polypeptides, 20 ff. 
requirements of, 5 ff. 
weight-pattern determination in, 15 

Creatinine clearance 
endogenous, 214 ff. 
applications, 217 f. 
blood specimen, 214 f. 
calculation, 216 f. 
dietary preparation, 214 
limitations, 217 
normal values, 217 
serum-creatinine concentration 
determination, 215 f. 
urine-creatinine concentration 
determination, 216 
urine specimen, 216 
exogenous—“undisturbed” method 
for (in rats), 244 ff. 

Crj’oscope, 194 

Curves, theoretical: calculation of, 
in countercurrent distribution, 
13 ff. 


11-Desoxycorticosterone acetate: in 
establishment of hypertension 
in rats, 261 ff. 

Deuterium: in total body water 
measurement, 168 ff. 

Dialj’sis: in electrophoresis, 90 

Dicker procedure: in determination 
of renal function (in rats) 
246 f. 

Diffusion: in specific precipitation in 
gels, 335 ff. 

double, see Diffusion, double 
simple, see Diffusion, simple 
special identification procedures, 
375 f. ’ 

theory of, 337 f. 

Diffusion, double, 359 ff. 
bacteriologic application of, 374 f 
defined, 359 
error sources, 372 f. 
identification in, 373 
in one dimension, 359 
from opposite parallel sources 
coalescence of zones, 370 f. 


reacting antigen determination, 
363 f. 

results, 362 f. 
technique, 360 f. 
theory, 361 f. 

from perpendicular sources 
practical considerations, 368 ff. 
techniques, 364 f. 
theory, 365 ff. 

sensitivity of technique, 371 f. 
in two tlimensions, 360 ff. 

Diffusion, simple, 339 ff. 

bacteriologic application, 374 
in one dimension, 339 ff. 

complex j^recipitating sj'^stems, 

352 f. 

component identification, 356 ff. 

error sources, 355 f. 

multiple precipitating systems, 

353 f. 

simple i)recipitating systems, 346 


technique, 340 ff. 

—sensitivity, 354 f. 
theory, 342 ff. 
in two tlimensions, 359 
Diffusion coefficients: in simple pre¬ 
cipitating systems in gels, 350 

Diodone: excretion of, 141 ff. 

renal clearance (in rats), 248 
renal transport (in rabbit kidney 
slices), 2.32 f. 

Diphtheria toxin-antitoxin reaction, 
326 ff. 

Diuresis, osmotic, see Osmotic diu¬ 

Diuretic substances in urine: bioas¬ 

with dogs, 207 ff. 
with rats, 206 f. 


Electrodes: for electrophoresis, 66 
96 ff. ’ ’ 

Electrolyte excretion 

during osmotic diuresis (in hydro- 
penic man), 196 ff. 
study of (in dog), 175 ff. 
calculations, 178 
e>^li;iwellular fluid volume study, 

—fluid expansion, i)rocedure for, 
181 tf. 

—osmotic diuresis, 183 ff. 
freezing point determination, 176 

procedures, 175 ff. 



Electron microscopy: prepiinition of 
tissue sections for, 234 ff. 

Electrophoresis, 63-106 
conductivity measurement in 
bridges for, 91 ff, 
cell for, 90 f. 
convection control in 
and channel width, 93 f. 
convective circulation, 93 
and Joule heat, 94 ff. 
and temjK'iature, 9(3 
definition, 03 
electrodes for, 00, 90 tf. 
ideal, 85 f., 99 
interference methods for 
evaluation of, 80 
Jamin fringes, 78 f. 

Rayleigh fringes, 70 tf. 
choice of cells for, 102 f. 
commercial equipment for, 101, 

resolution in, 101 f., 103 
pattern analysis 
areas for 80 f. 
channel cross-section in, 83 
channel depth in, 81 
enlargement factors in, 82 
fringe analysis in, 77, 83 tf. 
interpretation of, 85 ff. 
mobility computation, 82 f. 
optical distance in, 81 f. 
pattern interpretation in 

apparent and tiue composition, 


ideal situation, 85 f. 
motility vs. pH in, 86 f. 
preparative, 65, 66, 98 ff. 
j)rinciples of, 63 f. 
schlieren methods 
diaphragms, 74 
light sources, 73 f. 
limitations of 72 f. 
optical principles, 69 ff. 
optics of, 74 ff. 
scanning ])rocess, 711. 
solutions for 
Imffers, 87 ff. 

dialysis in preparation of, 90 
stock, 89 

in Tiselius cell, 64 ff. 

Encephalomyelitis virus, 315 

Enzymes . 

analysis by i)ai)er chromatography, 


])recipitin reactions, 316 ff. 

Escherichia coli 

analysis of H phage preparat ions ol, 

immunologic study of /t-gala(4o- 
sidase of, 311,319 ff. 

Esters: analysis by paper chromatog¬ 
raphy, 44 

Ethanol fractionation methods: for 
separation of antibody-active 
jiroteins from animal sera, 284 

alcohol for, 285 f. 

7 -globulins, 290 f. 
initial precipitation in, 290 
ionic, strength in, 287 f. 
pH in, 286 
procedures, 289 ff. 
protein concentration in, 288 f 
sulifractionation procedures 
chicken, 298 f. 
cow, 296 ff. 
dog, 293 
goat, 293 
guinea pig, 293 
horse, 294 ff. 
man, 293 f. 
pig, 298 

rabbit, 291 ff., 293 
rat, 293 

temperature in, 286 f. 

lOxcretion constants 
ofdiodone, 141 ff. 
of inulin, 140 f. 
of PAH, 141 ff. 

Extracellular fluid volume: study of, 
159 ff., 178 ff. 

analytical jirocedures, 163 ff. 
calibrated infusion technique, 161 
ff., 165 f. 

constant infusion-urine collection 
technique, 1(30 f., 165 
fluid expansion, procedure for, 181 

increased volume, 165 ff. 

and osmotic diuresis, 183 ff., 188 


hNtraction, see C’ountercurrent dis¬ 


Filtration rate and Tiupah measure¬ 
ment (in rats), 242 ff. 

Flocculation reaction, 323 ff. 

Fractionation, see Countercurrent 
distribution and Ethanol frac¬ 
tionation methods 

Freezing point 

depression—determination during 

osmotic diuresis, 194 
determination in electrolyte excre¬ 
tion study (in dog), 176 ff. 

Friedman procedure: for determina¬ 
tion of renal function in rats, 
247 f. 




/3-galactosidaso: immunologic meth¬ 
ods of study, 311, 31!) ff. 

Gels, specific precipitation in, 335- 

hacteriologically applical)le tech¬ 
niques, 374 f. 
double diffusion, 35!) ff. 

see also Diffusion, double 
simple diffusion, 33!) ff. 

see also Diffusion, simple 
special identification procedures, 
375 f. 

theoretical background, 337 ff. 

Glucosamine and precipitin reac¬ 
tions, 318 

Gravitational fields: sedimentation 
in, 107 

Growth factors and paper chroma¬ 
tography, 44 

excretion constant, 140 f. 

in infusion technique for renal-func¬ 
tion study, 157 f. 

in study of extracellular fluid in 
bo(lv water comj)artments, 100 
f., 163 ff. 

Ionic strength: during ethanol frac¬ 
tionation, 287 f. 


in body water (total) measurement, 
108 ff. 

in extracellular fluid measurement, 
107 f. 


Jamin fringes, 78 f. 

Joule heat: convection from, and 
electrophoresis, !)4 ff. 



Hemocyanin, 325 


in establishmenti of experimental 
hypertension (in rat-s), 261 ff. 
precijutin reactions of, 310 ff. 

Hydrogen ion concentration, see 

I lydropenic man: osmotic diuresis for 
determination of water and 
electrolyte excretion in, 192 ff. 

Hypertension: establishment in rats 

with hormones, 261 ff. 
by perinephritis, 258 f. 
with renal artery clamp, 259 ff. 
with saline, 263 


Immunochemical methods for deter¬ 
mining homogeneity of pro¬ 
teins and polysaccharides, 208- 

antibody j)roduction in exjjcrimen- 
tal animals, 271 ff. 

ethanol fractionation technioues 
284 ff. * ’ 

precipitin reaction, quantitative 
HI gels, 335 ff. 
in liquid media, 301 ff. 

Infra-red spectroscopy: in paper 
chromatography, 47 

Inlusion technique: in renal-function 
measurement, 156 ff 


clearance, in determining 
function, 246 ff. 


Kidne.v, .sec also Rimal blood flow, 
function, etc. 

osmotic limitations—measurement 
of, 194 ff. 

tubule excretion, in vitro study, 228 


/3-lactoglobulin, 314, 315 
Lanthanum nitrate: in tissue fixa¬ 
tion for electron microscopy, 
235 ff. 

Liesegang phenomenon, 350 


Mannitol: in osmotic diuresis, 192 

ff. ’ 

Manometer, microphonic: for in¬ 
direct determination of sys¬ 
tolic blood pressure in rats, 251 


P'fl^prfJ'hromatography in study of, 

water and electrolyte, 155 ff 
Micro-electrophoresis, see IClectro- 


Nephrotic syndrome: induction by 
anti-kidney serum, 264 ff. 


Osmic acid: in tissue fixation for 
electron microscopy, 235 ff. 



Osmotic diuresis 

for determination of water and 
electrolyte excretion in hydro- 
penic man, 192 ff. 
in dog 

procedure for, 188 flf. 

quantitative description, 183 ff. 
kidney in 

osmotic activity—determination, 

osmotic limitations—measure¬ 

ment, 194 ff. 

Ovalbumin, chicken: reaction of, with 
homologous rabbit antiserum, 
346 ff. 

1 ‘ 


clearance in determining renal func¬ 
tion (in rats), 247 
excretion constant, 141 ff. 
in infusion technique for renal- 
function stud}', 157 f. 
transport in study of tubular excre¬ 
tion in rabbit kidney slices, 
231 f. 

I’artition ratios 
of amino acids, 20 
definition, 5 

Peptides: analysis by paper chroma- 
tograijhy, 31 f. 

Perinei)hritis: in establishment of 
hypertension in rats, 258 f. 

pH: during ethanol fractionation, 
286, 288 

Phenol red transport for study of 
renal tubular excretion 
frog kidney slices, 230 
guinea pig kidney slices, 230 f. 
isolated fish tubules, 228 ff. 
rabbit kidney slices, 233 

Philpot-Svensson ojfiical system, 114 

Plasma disap[)earance rates: in renal 
function studies, 135 IT. 

Polypeptides: analysis by counter- 
curnmt distribution, 20 ff. 

Polysaccharide antigens: injection of, 
276 f. 

Polysaccharides: determination of 

homogeneity by immunochem¬ 
ical methods, 268-378 

Potassium dichromate: in tissue fixa¬ 
tion for electron microscopy, 
235 ff. 

Precipitating systems in gels 
classification, 339 
complex, 352 f. 
multiple, 353 f. 
simple, 346 ff. 

Precipitation, specific: theory of, 339 
Precipitin reaction (quantitative) in 
gels, 335 ff. 

bacteriologically applicable tech¬ 
niques, 374 f. 
double diffusion 
defined, 359 
error sources, 372 f. 
identification, 373 
in one dimension, 359 
from opposite parallel sources, 
360 ff. 

sensitivity of technique, 371 f. 
in two dimensions, 360 ff. 
from two perpendicular sources, 
364 ff. 

simple diffusion 

in one dimension, 339 ff. 
in two dimensions, 359 
special identification procedures, 
375 f. 

theoretical background, 337 ff. 
Precipitin reaction (quantitative) in 
liquid media, 301 ff. 
analysis of, 308 ff. 
antigen-antibodv mixture setting¬ 
up, 302 ff.‘ 

flocculation reaction, 323 ff. 
materials and methods, 301 ff. 
quantity determination of specific 
precipitates, 305 ff. 
tests of supernatants, 307 f. 
washing specific precipitates, 304 f. 

analysis by ultracentrifugation, 126 
antibody-active—separation from 
animal sera by ethanol frac¬ 
tionation, 284 ff. 
biosynthesis—analysis of, 319 ff. 
determination of homogeneity by 
immunochemical methods, 268 

synthesis—immunologic methods 
of study, 319 ff. 


Rayleigh fringes, 76 ff. 

Regaud’s fluid: in tissue fixjition for 
electron microscopy, 236 f. 
Renal blood flow measurement, 147 


direct measurement (rotameter), 
150 ff. 

extraction and clearance method, 

147 ff. 

Renal clearances, see Excretion con¬ 


Renal function, 

estimation of, 134 


Renal function (cont.) 
arginine Tm, microbiologic estima¬ 
tion (in dog), 220 ff. 
bioassay of antidiuretic and diu¬ 
retic substances, 201 ff. 
creatinine clearance 
endogenous, 214 ff. 
“undisturbed” method (in rat), 
244 ff. 

Diodrast clearance (in rat), 248 
electrolyte excretion studies (in 
dog), 175 ff. 

extracellular fluid volume study, 
178 ff. 

general procedures, 175 ff. 
osmotic diuresis, 183 ff. 
electron microscopy in, 234 ff. 
evaluation of methods, 248 f. 
filtration rate and Tiupah measure¬ 
ment (in rat), 242 ff. 
infusion technique, 156 ff. 
inulin clearance, 246 ff. 
during osmotic diuresis in hytlro- 
penic man, 192 ff. 

PAH clearance, 247 
by plasma disappearance rates, 135 
kinetics of excretion of injected 
substances, 135 ff. 
single injection technique, 139 ff. 
by space techniques, see Body 
water compartments 
in vitro methods for tubular excre¬ 
tion studies, 228 ff. 

Renal pathology, experimental 
arterial pressure measurement (in 

indirect recording, 253 ff. 
with microphonic manometer, 
251 ff. 

hypertension establishment (in 

with hormones, 261 ff. 
by perinephritis, 258 f. 
with renal artery clamp, 259 ff. 
with sodium chloride, 263 
nephrotic syndrome induction by 
anti-kidney serum, 2(54 ff. 
analysis (in rat), 265 f. 
antisera preparation, 264 f. 

Rotameter: in renal blood-How mea.s- 
urements, 150 ff. 


Scarlet fever toxin, .324 f. 

Schlieren methods in electrophoresis. 


analytical ultraccntrifuge for, 110 

ff. ’ 


mathematics of, 119 ff. 
optical methods of recording 
absorption method, 119 
cylindrical lens procedure, 114 
modified scale method, 118 f. 

Sedimentation constant, 109 f., 119 

Sera, animal: separation of antibody- 
active proteins from, 284 ff. 
see also Ethanol fractionation 

Serum albumins, 325 

Sodium chloride: in establishment of 
experimental hypertension in 
rats, 263 

Solutions: for electrophoresis, 87 ff. 

Solvents: for ascending paper chro¬ 
matography, 35 ff. 

Space techniques, see Body water 

Spectroscopy in paper chromatog¬ 
raphy, 47 

Sucrose in study of extracellular fluid 
in body water compartments, 
161 ff., 163 ff. 

Sugars: analysis by paper chroma- 
tograjjhy, 35 ff., 42 ff., 50 f. 



control in electrophoresis, 96 
during ethanol fractionation, 286 f. 

Tetanus toxin, 324 f. 

Thermistor: in electrolyte excretion 
study, 176 

Tissue sections: preparation for elec¬ 
tron microscopy, 234 ff. 

Toxins: precipitin reactions of, 316 


Tracers: ^in paper chromatography. 

Tritium: in total body w’ater meas¬ 
urement, 168 ff. 

Tyrosinase-antityrosinase reaction, 



Ultracentrifugation, 107 ff. 
accuracy of, 125 
equipment, 110 ff. 
interpretation of results, 125 f. 
ii^^^fJ|6matical relationships in, 

optical methods for 
—absorption method, 119 

cylindrical lens procedure, 114 


—modified scale method, 118 f. 



Ultracpntrifugiition (cont.) 
procedure during, 122 ff. 
uses for, 126 

preparative, apparatus for, 126 ff. 
sedimentation characteristics dur¬ 
ing, 107 f. 

analytical, 110 ff. 
preparative, 126 ff. 


antidiuretics in 
bioassay, 201 ff. 

—with dogs, 202 f. 

•—with rats, 201 f. 
concentration of, 204 f. 
creatinine concentration determi¬ 
nation, 216 

diuretics in—bioassay, 206 ff. 
with dogs, 207 ff. 
with rats, 206 f. 

Viruses: precipitin reactions of, 316 


Water, see also Body watt'r and Ex¬ 
tracellular fluid 

and electrolyte excretion in hydro- 
penia, 192 ff. 

intracellular, estimation of, 171 ff. 

Water and electrolyte metabolism, 
155 ff. 

see also Renal function, estimation 

Weight patterns: determination in 
countercurrent distribution, 15 


X-raj' diffraction: in pap('r chroma- 
tograj)hy, 47 


[Page numbers printed in i)old face indicate original contributions to this volume.] 


Aas, K., 218 
Abrams, A., 334 
.\l)rams, M., 259 
Acher, R., 54 
Adair, M. E., 332 
.\dams, M. H., 311, 332 
.\ddis, T., 214, 217, 218 
Agar, H. D., 105 
.\gner, K., 330, 335 
Ahrens, E. H., Jr., 22, 23 
Alberty, R. A., 104, 299 
.\lperin, L. J., 207 
Alpert, L. K., 233 
Alving, A. S., 145 
Ames, R., 212 
Anderson, R. S., 132 
Anderson, T. F., 240 
Anfinsen, C. B., 331 
Anselmino, K. J., 212 
Anslow, W. P., Jr., 191 
Antweiler, H. J., 104 
Archibald, R. C., 23 
Arden, T. V., 29, 54 
Argobast, R., 318, 333 
.\rm.strong, S. H., Jr., 104 
.\ronoff, 8., 54 
Auclair, J. L., 41, 46, 54, 00 
Awapara, J., 54, 58 


Baeher, J. E., 22 
Badger, J. S., 335 
Baer, H., 332 
Bailey, E. D., 110, 132 
Baker, R. F., 237, 239, 240 
Baldwin, I). S., 217, 219 
Baldwin, R. L., 133 
Baptist, V. H., 55 
Barbour, II. G., 331 
Barker, H. A., 37, 58, 00 
Barker, 11. G., 149 
Barnafi, L., 204, 205 
Barnett, II. L., 145 
Barry, G. T., 22, 23 
Barton-Wright, E. C’., 54 
Ba.ssham, J. A., 27, 54 
Bate-Smith, E. C., 28 54 
Batters by. A., 23 ’ 

Beams, j. W., 131 

Bechhold, H., 335, 370 

Becker, E. L., 337, 344, 350, 376, 377 

Behan, M. A., 270, 281 

Bendich, A., 333 

Benedict, F. G., 210, 212 

Benson, A. A., 27, 54, 55, 61 

Bentlev, H. R., 54 

Berg, C. P., 59 

Berger, E. Y., 150, 157, 159 

Berliner, R. W., 150, 159 

Bernhard, W., 240 

Beyer, Karl II., 220, 220, 230, 231, 

Bezer, A. E., 333 
Bidwell, PI, 334 
Birnie, J. H., 209, 211, 212 
Blackburn, S., 55 
Blegen, p]., 218 
Blix, Gunnar, 131 
Block, R. J., 02 
Boldingh, J., 52, 55 
Bonanto, M. V., 281 
Bonner, W. ()., 331 
Bonsnes, R. W., 215, 218 
Bordley, J., Ill, 145 
Borsook, H., 55 
Borysko, E., 237, 240 
Boscott, R. J., 52, 55 
Bott, P. A., 220 
Botty, M. C., 55 
Bowen, H. E., 330, 335 
Boyd, W. C., 334 
Bradlej', Stanley PI, 147, 149 
Brandt, W. L., 203 
Brante, G., 41, 55 
Bninting, B. P^., 331 
Bnitton, A. G., 232, 233 
Brattsten, I., 100 
Braun-Menendez, Pk, 244, 249 
Bretschneider, L. II., 239 
Briggs, D. R., 100 
Briggs, F. N., 01 
Brod, J., 215, 218 
Brodie, B. B., 173, 174 
Brodsky, William A., 155, 150, 191, 
192, 195, 199, 200 
Bronsted, J. N., 87, 104 
Brown, A., 132 
Brown, D. M., 300, 334 
Browm, P\, 37, 44, 55 
Brown, R., 336, 355, 376 



Brown, R. J., 31 
Bryson, J. L., 55 
Budka, M. J. E., 104 
Buerger, E. Y., 173 
Bull, H. B., 41, 55 
Bullock, L. T., 212 
Bulky, E. L., 281 
Burn, J. H., 201, 204, 205 
Burnett, F. M., 281 
Burstall, F. H., 55 
Burton, C. J., 55 
Burton, K., 281 
Burton, R. B., 55, 02 
Bush, M. T., 12, 23 
Byroin, F. B., 212 

153, 155, 218, 233, 239, 242, 244, 

246, 248, 249, 250, 257, 258, 261, 

202, 263 . . , 

Cornish, R. E., 21, 23 
Corsa, L., Jr., 173 
Cosslett, V. E., 240 
Cournand, A. F., 147, 149 
Craig, Lyman C., 1, 2, 3, 22, 23, 24, 

Crammer, J. L., 50 
Cross, R. J., 231, 233 
Croxatto, H., 204, 205 
Crumpler, H. R., 47, 50 
Culbertson, J. T., 314, 333 
Curnen, E. C., 131 



Calvin, M., 55, 58, 01 
Cdmara, Augusto A., 214, 218 
Cammarat.a, F. S., 299 
Campbell, J. R., 43, 40, 59 
Campillo, A., 333 
Cann, J. R., 131, 280, 299 
Cannan, R. K., 58 
Canti, R. G., 241 
Carlinfanti, E., 272, 281 
Caspary, E. A., 335, 378 
Cassidy, H. G., 35, 58 
Cavallini, D., 55 
Cesari, E., 377 
Chaikoff, I. L., 01 
Chambers, Robert, 228 
Chargaff, E., 42, 55, 01 
Charney, J., 282 
Chasis, H., 139, 140, 218 
Chiodi, H., 244, 249 
Chow, B. F., 333 
Christ, C. L., 47, 55 
Chung, S. F., 335, 374, 378 
Ciminera, J. L., 220 
Clark, J. K., 148 
Claude, A., 240 
Clegg, D. L., 35, 55, 59 
Cohen, S., 318, 333 
Cohen, S. M., 330, 370 
Cohen, S. S., 55, 00 
Cohn, 27 
Cohn, C., 02, 153 
Cohn, E. J., 284, 299 
C'ohn, Melvin, 268, 271, 281, 283, 299, 
300, 301, 315, 322, 320, 331, 332, 
334, 339, 352, 375, 370 

Conn, 149 

Consden, R., 25, 20, 27, 28, 29, 30, 
35, 30, 38, 55, 50, 104 
Cook, A. II., 50 
Cook, G. II., 70, 105 
Cooper, C., Jr., 209 
Corcoran, A. C., 134, 139, 145, 149, 

Dalton, A. J., 234, 239, 240 
Danysz, J., 303, 331 
Davidson, C. S., 281, 334 
Dawson, C., 333 
Day, R. L., 332 
Dean, H. R., 332, 330 
Deane, Norman, 155, 156, 159, 172, 

Dearborn, E. H., 233 
Debains, E., 377 
Densen, P., 12, 23 

Dent, C. E., 31, 32, 30, 38, 40, 41, 45, 
40, 47, 49, 50, 01 

Deutsch, H. F., 131, 281, 283, 284, 
299, 300, 301, 322, 332 
Devanev, VV. A., 02 
Dicker, S. E., 211, 212, 244, 240, 247, 

DiLapi, M. M., 282 

Dillon, J. F., 131 

Dimler, R. J., 57 

Doak, B. W., 332 

Dole, V. P., 80, 104 

Dolliver, M. A., 23 

Dominguez, Rafael, 135, 139, 145,155 

Donnan, F. G., 104 

Drummond, D. G., 240 

Drury, D. R., 203 

Dubos, R., 219 

Duclaux, J., 347, 370 

Dunn, M. S., 220 

Dunning, M. F., 173 

Dunum, E. L., 104 


Earle, D. P., Jr., 150, 157, 158, 159, 

Edelman, I. S., 109, 172, 173, D4 
Eden, M., 240 
Edman, P., 50 
Eidinoff, M. L., 170, 173 
Eigen, E., 44, 02 



Eisen, H. N., 331 

Elek, S. D., 337, 360, 364, 365, 369, 
371, 373, 375, 376 
Elliott, C. H., 281 
Elliott, H., 131 
Ellis, M. E., 205 
Elsden, S. R., 56 
Engel, L. C., 333 
England, 27 
Evans, H. M., 23, 132 
Eversole, 209 
Ewart, R. II., 133 


Farber, S., 226 
Farber, S. .1., 156, 157, 159 
Farr, L. E., 204, 264, 267 
Feigl, F., 56 
Felsing, W. A., 105 
Fenichel, R. L., 105 
Fenner, F., 281 
Fenton, E. L., 335, 378 
Ferguson, M. H., 145 
Fink, K., 56 
I4nk, R. M., 56 
Finkelstein, N., 227 
Fisher, R. B., 51, 56 
Fisher, S., 226 
Flexner, L. B., 155, 173 
Flood, A. E., 50, 51, 56, 57 
Flore}^, K., 61 
Folin, 0., 204, 226 
Folk, B. P., 191 
Fontaine, T. D., 58 
Foota, R. W., 267 
Forbes, G. B., 173 
Foreman, E. M., 60 
Forster, Roy P,, 228, 230, 233 
Forsyth, W. G. C., 57 
Fowden, L., 49, 50, 52, 57 
Fredericq, E., 131 
Freed, C. F., 257 
Freed, S. Charles, 251 
Freinkel, N., 174 

Freund, J., 271, 275, 276, 281, 282, 

Fried, J., 23, 24 
Friedel, R. A., 23 
Friedman, C., 249 
Friedman, Meyer, 251, 257 
Friedman, S. J., 247, 248, 249 
IViedwald, W, F., 131 
Fromageot, C., 54 
Frontali, N., 55 
Fullam, E. F., 240 


Ganguli, M., 226 
Gass, S. R., 226 
Gaudino, M., 173 

Gaunt, Robert, 155, 209, 212 
Gautier, A., 240 
Gengras, R., 249 
Geren, B. B., 240 
Gessler, A. E., 240 
Gettner, M. E., 240 
Gilkey, C., 267 
Gilman, A., 201, 204 
Ginsburg, M., 210, 211, 212 
Giri, K. V., 44, 57 
Gitlin, D., 278, 281, 334 
Goebel, W. F., 319, 333 
Gofman, J. W., 126, 131 
Goldblatt, H., 139, 261 
Goldring, W., 139, 146 
Golumbic, C., 21, 23 
Golumbic, E., 23 
Goodall, R. R., 57 
Goodban, A. E., 61 
Goodman, A., 201, 204 
Goodner, K., 281 

Gordon, A. H., 25, 49, 55, 56, 57, 

Gorini, L., 282 
Gosting, L. J., 104, 299 
Grabar, P., 276, 278, 282, 325, 332, 

Graubarth, 195 
Green, C., 55 

Greenspon, S. A., 264, 267 
Gregg, Donald E., 259, 261 
Gregory, J. Delafield, 2, 22 
Grev, C. E., 240 
Griffith, J. Q., Jr., 209, 212, 213 
Grollman, Arthur, 155, 204, 205, 206, 
250, 257 

Grubhofer, N., 23 
Gutter, F. J., 125, 131 


Hackel, D., 267 
Hadidian, Z., 209 
Hahn, J. \V., 55 
Hald, P. M., 191 
Hall, C. E., 263 
Ham, G. C., 201, 204 
Hamilton, AV. F., 331 
Hance, R. T., 240 
Hanes, C. S., 35, 40, 57 
Hanks, J. H., 336, 376 
Hanson, E. M., 104 
Hare, A., 202, 204 
Hare, K., 212, 218 
Hare, R. S., 218 
Harfenist, E. J., 23 
Harned, H. S., 104 
Harris, G., 54 
Harrison, T. R., 250, 257 
Hartmann, J. F., 240 
Hastings, A. B., 174, 191 



Haugen, H. N., 218 
Ilaurowitz, F., 332 
Hausmann, W., 23, 36, 57 
Hawthorne, J. R., 50, 51, 57 
Hecht, S., 131 
Hecker, E., 23 
Hedenius, A., 132 
Hedges, E. S., 376 

Heidelberger, M., 276, 282, 286, 300, 
309, 312, 331, 332, 333, 334, 335, 
336, 377 

Heller, H., 210, 211, 212, 249 

Hennessy, J. N., 334 

Herriott, T. M., 282 

Hess, E. L., 300 

Hestrin, S., 333 

Hevesy, G., 168, 173 

Heymann, Walter, 264, 266, 267 

High, L. M., 133 

Hill, R. F., 267 

Hiller, A., 175, 191 

Hillier, J., 240 

Hirst, E. L., 46, 51, 56, 57 

Hitchcock, i)avid I., 104, 376 

Hodes, P. J., 213 

Hofer, G., 168, 173 

Hoffman, F., 212 

Hogehooin, G. H., 131 

Holiday, E. R., 57 

Holmherg, C. G., 131 

Holt, L. B., 274, 276, 282 

Hooker, S. B., 334 

Horn, D. B., 146 

Horne, R. E., 42, 57 

Horsfall, F. L., 131 

Hotchkiss, R. D., 51, 57 

Hottle, G. A., 334 

Howley, E. ^I., 263 

Hulton, E., 282 

Hunter, T. H., 331 

Hurwitz, D., 212 

Hutchison, D., 24 


Isherwood, F. A., 35, 37, 38, 40, 57, 


.Jacobson, li., 57 
Jager, B. V., 322, 334 
.James, A. H., 173 
J.ames, \V. G., 57 
.Jeanes, A., 57 
.Jenkins, 209 
Jen.sen, G., 282 
.Jermyn, M. A., 37, 38, 58 
.Jerne, N. Iv., 328, 335 
.Johlin, .J. M., 17(), 191 
Johnson, E. A., 57 

.Johnson, P., 99, 104 
Johnston, J. P., 123, 131 
Jolliffe, N., 218, 226 
Jones, J. K. N., 46, 51, .56, 57 
Jones, T. S. G., 29, 32, 35, 46, 52, 

Joslyn, M. A., 30, 58 
Jutisz, M., 54 


Ivabat, E. A., 80, 104, 106, 270, 277, 
279, 280, 282, 301, .304, 308, 318, 
322, 331, 332, 333, 334, 335 
Kahler, H., 240 
Ivahn, D. S., 68, 104 
Kaplan, S. A., 200 
Ivarlson, V. P., 23 
Katz, B., 153 
Kaufman, D., 86, 105 
Ivedwick, P. A., 270 
Kegeles, G., 104, 125, 131, 3.34 
Iveith, N. M., 174 
Ivelsch, J. J., 240 
Kempf, G. F., 2.50, 257 
Kendall, F. E., 282, 309, 312, 314, 
.3.31, 332, 33.3, 336 
Ivendall, F. F., 326, 335 
Ivennedv, E. P., .37, ,58 
Kent, \\ W., .59 
Jverposar, A., 209 

Iversten, H., 255, 2.57, 260, 261, 266, 

Keston, A. 8., .50, 58, 169, 173 
Keutmann, E. H., .5.5, 62 
Kilvington, T. B., 201, 204, 212 
Kinell, P., 123, 125, 131 
King, E. Ij., 104 
Kirby, A. H. M., 104 
Kirby, H., 32, 62 
Ivirkbride, M. B., 336, 376 
Ivlett, 93 
Klisiecki, A., 200 
Knaub, V., 332 
Knight, B. C. J. G., .333 
Ivnoll, J. E., 170, 17.3 
Knowlton, Abbie I., 266 
Kolinsky, M., 153 
Konige, H. B., 24 
Kopp, II., 3.35, 378 
Korkes, S., 333 
Kostif, J. V., 44, .58 
Kotjitko, J., 21.5, 218 
Kowkabany, G. N., 35, 58 
Kozloff, \j. M., 132 
Krakower, (k A., 264, 267 
Krieger, V. I., 201, 204, 212 
Kritchevsky, I)., .58 
Kritclievsky, T. H., 53, 58 
Jvrueger, R. C., 300 



Kubv, S. A., 23 
Kunkel, II. G., 331 


Labhart, II., 105 
Ladd, M., 151), 173, 180, 191 
Lamanna, C'., 332 
Lamm, 119 

Landis, E. M., 201, 204 
Landstniner, K., 271, 282, 333 
Landua, A. J., .58 
Lapresle, C., 370 
Lardy, H. A., 23 
Latta, Harrison, 239, 240 
Lauffor, M. A., 132 
Leboau, L., 370 
L(“ Breton, E., 333 
Le.Metayer, 352 
Levi, A. .4., 57 
Levine, C., 55 
Levitt, G., 212 
Levitt, M. F., 173 
Levkoff, 195 
Levy, M., 58 
Lewis, Lena A., 104, 131 
Li, C. H., 132 
Libby, K. L., .331 
Lieberman, S. V., 23 
Lilienthal, J. L., 191 
Lilly, J. C., 173 
Lindemann, W., 204, 207 
Lindgren, F. T., 131, 132 
Lippman, Richard W., 244, 249 
Lipschitz, W. L., 207, 209 
Little, J. .Maxwell, 1.55, 201, 204, 20.5, 
206, 209, 212 > . , 

Lloyd, C. W., 212 
Lobotsky, J., 212 
Lochte, H. L., 23 
Lockwood, L. B., 31 
Loi.seleur, J., 301, 331 
Long, C. L., 44, .58 
Longenecker, \V. IL, 32, 58 
Longsworth, Lewis (!., 63, 105 
Lotmar, \V'., 10.5 
Lowry, O. H., 331 
Luetscher, J. A., Jr., 105 
Lugg J. W. IL, 29, 37, 44, 58 
Lund, IL Z., 204, 207 
Lundgren, II. P., 133 


Ma, U. .M., 58 
.McCance, 199 
•McCarthy, J. L., 105 
.McCulloch, D., 240 
.McDougall, E. J., 174 
•MacFarlane, .M. G., 333 

McFarren, E. F., 38, 59 

Macheboeuf, M., 38, 59 

41aclnru*s, Duncan A., 104, 105 

McKinley, G. M., 240 

MacLeod, C. M., 333 

McOmie, J. F. W., 00 

MacPherson, C. F. C., 282, .331, .3.32 

Maegraith, B. G., 330, 377 

Maga.sanik, B., 58 

Malkiel, S., .3.32 

.Mallette, F. .M., 333 

Maloney, P. S., 3.34 

Markham, R., .58, 00 

Markley, 149 

Markowitz, 11., 3.3.3 

Manshak, .4., .58 

Marshall, E. K., Jr., 200, 232, 2.33 
.Martel, F., 244, 249 
.Martin, A. J. 1’., 2.5, 29, 38, .50, .55, 50, 
57, .59, 104 
Marton, C4, 240 
Marvin, H. IL, 104 
IVIarx, IL, 212, 213 
Mason, M., 59 

.Masson, G. .M. C., 242, 244, 258, 
261, 2()2, 263 
Masugi, IM., 204, 2()7 
.Maurer, P. 11., 3.32 
.May, .M., 105 

Mayer, M., 104, 3.31, 3.34, .335, .377 
.Mayer, M. M., 270, 277, 279, 282, 
.301, 308, 331 
Mazur, .4., .33.3 
Mazurek, C., .334 
Melville, K. I., 213 
Meyer, H. W. H., 23 
MichaelLs, L., 105 
.Mighton, H., 23 
Miles, A. A., 283 
Miller, B. F., 145, 219 
Miller, F. K., 01 
Mitchell, T. J., 55 
.Mitchison, D. .4., .3.50, 377 
Mittelmann, R., 50, 59 
Moloney, J. P., .334 
Monod, J., 281, .332 
.Moore, 1). IL, 95, 10.5, 100, 212, 

.Moore, F. D., 172, 173, 174 
.Moore, S., 28, .50, 59, 01 
.Mora, T. P., 59 
.Morris, M. S., 104 
.Morrison, G. A., 50 
Morrison, K. C., 104 
Mo.sley, V. M., 00 
Moss, W. Glen, 252 
Mould, Y., 333 
Muller, R. H., 35, .59 
Munier, R., 38, 59 
.Munnel, E. R., 201 



Munoz, J., 337, 37G, 377 
Muri)hy, K. A., 23 
Murray, J. P., 334 


Naftalin, L., 59 
Nelson, N., U)4, 174 
Nelson, P. W., 180, 191 
Neuberger, A., 104 
Neuratli, II., 132 
Newman, E. V., 145 
Newman, S. B., 237, 240 
Niehol, J. C’., 281, 299, 300 
Nichols, J. H., 110, 132 
Nichols, P. L., 23 
Nickel, James P\, 147 
Nickerson, M., 334 
Nicolle, M., 336, 377 
Nier, A. O., 169, 174 
Norris, F. C., 43, 46, 59 


Oberling, C., 240 
O’Brien, H. C., 240 
Ochoa, S., 333 
Ogston, A. G., 123, 131 
O’Keefe, A. E., 21, 23 
Olbrich, O., 145 
Oliver, J., 249 

Olmsted, F'rederick, 253, 257 
Oncley, John Lawrence, 131, 132 
Orchin, M., 23, 

Ouchterlonv, O. 330, 335, 337, 359, 
360, 36L 362, 363, 364, 3()5, 368, 
369, 371, 373, 377 

Oudin, J., 271, 282, 314, 326, 335, 

Overell, B. T., 29, 37, 44, 58 
Owen, B. B., 104 
Owens, H. S., 61 

Pace, N., 168, 169, 174 
Pacsu, E., 41, 59 

Page, I. H., 139, 145, 250, 25/, 259, 
262, 263 

Painter, R. IL, 233 
Palade, G. E., 240 

Pappenheimer, A. M., Jr., 281, 282, 
323, 325, 326, 327, 328, 329, 331, 
334, 335 

Parsons, D. S., 56 

I’artridge, S. M., 36, 3/, 43, 5J, (>0 

Patton, A. R., 60 

Pauly, H., 42, 60 

Peace, D. C., 237, 239, 240 

Pedersen, K. O., 110, 119, 125, 131, 

Pendergrass, E. P., 213 
Peniston, Q. P., 105 
Perez, J. J., 334 
Perley, A., 173 
Perlman, E., 319, 333 
Perlmann, G. E., 86, 104, 105 
Petermann, M. L., 132 
Peters, J. P., 191 
Peters, T., Jr., 331 
Peterson, I). IL, 60 
Petri(>, G. F., 336, 359, 371, 374, 377, 

Phillips, I). M. P.,60 
Phillips, R. A., 149 
Philpot, J. St. L., 72, 76, 105, 114 
Pickels, Edward G., 72, 107, 110, 131, 

Pillemer, L., 335, 352 
Pitts, Robert F., 198, 225, 226 
Plant, G. W. E., 23 
Pollard, A. L., 42, 57 
Pollard, F. H., 60 
Polley, J. R., 249 
Poison, A., 49, 60, 68, 104 
Pomerene, E., 139 

Pope, C. G., 60, 326, 327, 328, 329, 
330, 335, 359, 378 
Post, O., 23 
Power, AI. IL, 174 
Prasad, A. L. N., 44, 57 
Pratesi, G., 283 
Pratt, A. W., 240 
Pratt, J. J., 41, 60 
Pressman, D., 267 
Prochdzka, 2., 41, 60 
Proom, H., 60 
Prudhomme, R. G., 332 
Putnam, F. W., 132 


Rdbek, V., 44, 58 
Ralli, E. P., 204, 205, 206 
Ramon, G., 282, 336 
Randall, R., 333 
Ranges, H. A., 149 
Rao, S. S., 334 

Rapoport, S., 191, 195, 199, 200 

Rau, G. C., 259 

Record, B. R., 270 

Reed, L. J., 60 

Reineke, L. M., 60 

Reiner, L., 335, 378 

Reiner, M., 105, 106 

von Rhorer, L., 200 

Richards, A. G., Jr., 240 

Richards, A. N., 226 

Riggs, B. C., 233 



Rimington, C., 60 
Rittenberg, D., 169, 173 
Ritter, E. R., 153 
Robertson, J. S., 159, 169, 173, 174 
Robinson, E. S., 323, 326, 327, 328, 
329, 331 

Robinson, F. H., 204 
Robson, James S., 135, 138, 139, 140, 
145 146 155 

Rodbai-d, Simon, 257, 259, 262, 263 

Roe, A. S., 333 

Rojas, G., 204, 205 

Ross, J. D., 131 

Rotlien, A., 132 

Rozsa, G., 241 

Russo, H. F., 225, 226 


Sakaguclii, S., 42, 60 
Salehar, AI., 267 
Sanger, F., 32, 60 
Sapirstein, L. A., 263 
Sato, Y., 22, 23 
Savat, R. S., 300 
Scatchard, G., 132 
Schachman, H. K., 132 
Schachter, D., 174 
Scherp, H. W., 314, 333 
Scldoerb, P. R., 174 
Schmidt, C. E. A., 60 
Schneider, K., 213 
Schneider, W. C., 131 
Scdioch, Henry K., 214 
Schoenheimer, R., 169, 173 
Scholander, P. F., 306, 331 
Schreiner, G. E., 159, 164, 173, 174, 

Schwartz, I. L., 172, 173, 174 
Schwerin, P., 332 
Scofield, P. P\, 132 
Scott, D. B. AI., 55, 60 
Seastone, C. V., 282 
Seegal, Beatrice, C., 266 
Selkurt, Ewald E., 149, 150, 154, 157, 
159, 191 > . , , 

Sellers, A. I., 249 
Selye, H., 263 
Shannon, J. A., 226 
Shaw, AI., 333 
Shedlovsky, T., 90, 104, 105 
Shevky, AI. C., 266, 267 
Shipley, R. E., 152, 153, 154 
Shooter, E. AI., 99, 104 
Shorr, E., 333 
Sia, R. H., 335, 374, 378 
Sickles, G. M., 333 
Sihtola, H., 125, 132 
Silvette, H., 204 
Simpson, AI. E., 132 

Siri, W. E., 169, 174 
Sirota, J. H., 216, 217, 218, 219 
Sjostrand, F., 241 
Slessor, A., 211, 213 
Smadel, J. E., 264, 267 
Smith, E. L., 300, 334 
Smith, Homer, 155 
Smith, H. W., 139, 143, 146, 156, 173, 
174, 191, 22(), 227, 233, 242, 249 
Smith, J. 1)., 58, 60 
Smith, R. F., 10() 

Smith, \V. W., 143, 146, 227 
Smolens, J., 282 
Soberman, R., 170, 174 
Sobin, S. S., 257, 2.59 
Solomon, A. K., 169, 174 
Solomon, D. H., 264, 267 
Soloway, S., 169, 174 
Song, H., Jr., 331 
Spark, A. H., 62 
Spicer, C. C., 350, 377 
Spreull, E., 146 
Stadie, W. C., 233 
Stadtman, E. R., 60 
Stadtman, F., 61 
Stafford, D. D., 266, 267 
Stanier, R. Y., 31 
Stanley, W. AI., 132, 332 
Stark, J. B., 61 
Staub, A. AI., 325, 332, 334 
Staub, H., 105 
Steabben, D., 336, 378 
Steele, J. AI., 173 
Stein, W. H., 28, 50, .59, 61 
Steinitz, K., 219 
Stene, S., 23 

Stepka, VVilliiim, 25, 30, 36, .56, .58, 

Stern, K. G., 106 
Stevens, 11. AI., 60 
Stevens, AI. F., 60, 335, 378 
Steward, F. C., 32, 36, 4.5, 46, 56, 61 
Stewart, C. P., 145, 146 
Stiller, E. T., 23 
Stodola, F. H., 31 
Stokes, J. L., 223, 227 
Strangeways, T. S. P., 241 
Study, R. S., 154 
Sumner, J. B., 282 
Sunderstrom-Song, K., 331 
Svedberg, T., 108, 110, 114, 119, 125, 

Svensso^i^^H., 72, 76, 103, 106, 114, 

Swart, E. A., 24 

Swerdlow, AI., 237, 240 

Swift, H. F., 264, 267 

Synp, R. L. AI., 25, 27, 46, 57, 59, 




Tahoury, F. J., 378 
Taboury, M. F., 378 
Taggart, John V., 228, 231, 233 
Takahashi, W. N., 01 
Taurog, A., 01 
Taussky, 11. H., 215, 218 
Taylor, Alice C., 104 
Taylor, A. R., 132 
Taylor, G. L., 332 
Theobald, G. W., 213 ' 

Thompson, J. F., 45, 50, 52, 01 
Thompson, J. O., 133 
Tiselius, A., 53, 58, 04, (iO, 72, 80, 100, 

Tishkoff, G. II.,01 
Titus, E., 23, 24 
Titus, E. J., 23 
Toennies, G., 02 
Toll, D., 335 
Tomarelli, R. M., 01 
Tomkins, R. G., 58 
Tong, W., 01 
Topley, 335 
Torriani, A. M., 334 
Toschi, G., 55 

Treffers, H. F., 100, 282, 333, 334 

Tristram, G. R., 01 

Troast, L., 220 

Tschesche, R., 24 

Turkand, H., 219 

Turner, Louis B., 155, 204 


Udenfriend, S., 58 
Uhlenhuth, Von S., 271, 283 
Umbarger, H. E., 58 
Urban, F. F., 210, 212 
Ureen, H. J., 249 


van Dyke, H. B., 212 

Van Heyidngc'ti, \V. E., 327, 334 

VanBlyke, D. I)., 175, 191 

Veil, S., 378 

V^erney, E. B., 200 

Vernon, L., 54 

Villarreal, H., 217, 219 

Vischer, E., 01 

Vogel, 11. J., 58 


Waksman, S. A., 24 
Wales, M., 124, 133 
Waley, S. G., 02 
Walker, A. M., 213 


Wallace, G. B., 174 
Walter, A. W., 281 
Wang, M., 249 
Ward, 8. M., 331 
Ward, W. II., 133 
Warner, R. C\, 325, 335 
Warren, J. V., 149 
Watson, J., ()2 
Weber, I., 325, 335, 330 
Webley, I). M., 57 
Weil, R., 314, 333 
Weir, E. G., 174 
Welt, L. N., 180, 191 
Wesson, Laurence G., Jr., 155, 150, 
175, 191, 199 

West, C'. I)., 191, 198, 200 
Westall, R. G., 28, 54 
Westfall, B. B., 220 
Wetter, L. R., 281, 283, 299, 322, 

Wetterlow, L. H., 281, 334 
Weygand, h'., 24 
White, H. L., 23 
White, J. U., 105 
Whitehead, J. K., 54 
Wiebelhaus, V. D., 233 
Wiedemann, E., 100 
Williams, J. R., Jr., 250, 257 
Williams, J. W., 133 
Williams, R. J., 32, 02 
Williams, W. L., 44, 58 
Williamson, B., 24 
Wilson, 335 

Wilson, C., 152, 154, 212 
Wilson, G. 8., 283 
Wilson, P. 0., 00 
Winegard, II. M., 42, 02 
Winkler, A. W., 219 
Winsten, W. A., 44, 02 
Winternitz, J., 145 
Wintersteiner, ()., 23 
Wise, C. 8., 57 
Woiwod, A. J. t50, 52, 00, 02 
Wolf, .V. V., 149, 155 
Wolfson, ^\'. Q., 02 
Wolter, IL, 72, 100 
Woods, B., 204, 200 
Woolfolk, E. ()., 23 
Work, E., 02 

Wright, Lemuel D., 220, 227 
Wu, H., 204, 220 
W'yekoff, R. W. G., 00, 241 


Zacharius, R. M., 01 
Zaffaroni, .A.., 53, 55, 02 
Zanco, M., 283 
Zierler, K. L., 191 
/,Ziff. M., 173 


30 R 

1 j 

! ' 




Il I 


Methods in medic.